Eyewitness Identification: Lineups, Showups, and Best Practices
Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy
Eyewitness identification is a formal attempt to determine whether a person seen during an event can be recognized later in a photo array, live lineup, or one-person showup. Its value depends on what the witness actually perceived, how the identification test was constructed, what the witness was told, and how the first decision was preserved. An identification can provide important investigative evidence, but it does not itself establish guilt, prove that the suspect committed the alleged act, or supply a numerical probability that an individual is guilty. The National Research Council’s comprehensive review and the American Psychology-Law Society’s policy recommendations provide the central scientific framework.
This guide explains what lineups and showups do, how investigators should protect identification evidence, why fair procedures matter, what accuracy research can and cannot establish, and how U.S. legal standards interact with psychological science. It distinguishes practices strongly supported by research from contested methodological choices. Legal admissibility, procedural rules, and required remedies depend on the applicable jurisdiction and the specific case.
What Is Eyewitness Identification?
In forensic psychology, eyewitness identification is a recognition-memory test with legal consequences. A witness compares a present person or image with memory of an earlier encounter. The task is susceptible to recognition errors because perception and memory are selective, reconstructive, and sensitive to conditions at encoding, storage, and retrieval. The witness’s sincerity is a separate question from whether the identification is accurate. Identifying the Culprit describes why perception, memory, police procedure, and courtroom evaluation must be considered together.
A suspect is the person investigators have reason to examine; a filler is a known-innocent individual included to test whether the suspect is recognized above plausible alternatives. A target-present test includes the actual perpetrator; a target-absent test does not. In laboratory research investigators know which condition applies. In a real case, whether the suspect is the perpetrator is typically the very question under investigation. This difference changes what can be measured directly and what must be inferred statistically. See the 2025 analysis of laboratory and Houston field evidence.
Photo Lineups, Live Lineups, and Showups
Photo arrays and live lineups
A photo array presents several photographs, commonly a suspect and multiple fillers, and asks whether the witness recognizes the perpetrator. A live lineup presents people physically. Both can be designed to permit a meaningful choice of “none of these people.” Photos are often more practical, whereas live procedures involve additional logistical constraints and may differ in posture, voice, lighting, or appearance. The National Institute of Justice’s eyewitness evidence guide covers construction and documentation of each format.
The quality of a lineup depends on its composition. If one person alone matches the witness’s prior description or appears markedly different in age, complexion, hairstyle, clothing, or image quality, selection may reflect the conspicuousness of that person rather than recognition from the event. Fair fillers are chosen to make the suspect blend into a plausible array while preserving diagnostic facial information. A 2025 NIJ-sponsored investigation of lineup composition specifically examined how fairness affects identification performance.
Showups: identifying one person in the field
A showup presents one suspect to one witness, frequently shortly after the event and near the scene. It offers speed and may help officers promptly release a person who is not recognized. It also exposes the witness to a highly suggestive situation: because only one person is shown, the witness may infer that investigators expect that individual to be identified. Police custody, handcuffs, visible officers, or statements implying that the person has been caught can compound that implication. The 2020 AP-LS recommendations advise avoiding showups when practicable and using safeguards when exigency makes them necessary.
A showup can result in a correct identification, an incorrect identification, or a nonidentification. The fact that it occurred shortly after the event does not immunize it against suggestion. A 2026 study of live showups found relationships between immediate confidence and accuracy in controlled and field settings, but the field study used an evidence-strength proxy rather than known ground truth; its results do not make any specific showup conclusive. Read the peer-reviewed study.
Simultaneous and sequential presentation
In a simultaneous lineup all faces are visible together. In a sequential lineup they appear one at a time. Earlier theories proposed that sequential presentation discourages relative comparison and therefore protects innocent suspects; subsequent signal-detection and receiver-operating-characteristic research has challenged the idea of a universal sequential advantage. The National Academies report characterized the evidence as unsettled, while a 2025 review combining laboratory and field analyses described results favoring simultaneous presentation in the particular datasets analyzed. Investigators should follow their jurisdiction’s validated, standardized protocol rather than claim either format is infallible.
Procedure details also matter: whether witnesses may revisit photos, whether a sequential presentation stops after a selection, whether all faces are shown, and the precise instruction wording can alter decisions. Comparisons of “simultaneous” with “sequential” are only interpretable when these details and the operating criterion are specified. The separate question of which format performs better belongs to a dedicated comparative review; here, the operational priority is that the chosen procedure be fair, blind, properly instructed, and fully recorded.
How a Proper Eyewitness Identification Procedure Works
1. Establish an independent investigative basis and preserve the initial account
Investigators should document why this particular person is a suspect before testing a witness’s memory. A lineup should serve as an evidence-gathering procedure, not as a way to manufacture support for an unsupported suspicion. Before a photo array or lineup, record the witness’s own description of the perpetrator and the event in a nonleading interview. Separate what the witness saw personally from information learned afterward through other witnesses, media, social networks, or investigators. The 2020 AP-LS guidance explicitly recommends evidence-based suspicion and a prelineup interview.
Important initial questions include the witness’s opportunity to view the face, approximate distance, lighting, duration of exposure, any obstruction or disguise, familiarity with the person, intervening encounters, and whether a description was offered before police named or displayed a suspect. Those details help interpret an eventual choice; they should not be converted into an improvised checklist claiming mathematical precision. Investigators should preserve spontaneous descriptions, uncertainty, and inconsistencies without coaching the witness to improve an account.
2. Construct one fair array with appropriate fillers
Select fillers primarily in relation to the witness’s description, while considering whether the suspect would stand out. A familiar practical convention is one suspect with at least five fillers, although exact composition requirements differ across agencies and jurisdictions. Avoid conspicuous differences in photo cropping, background, age, race-related appearance, facial hair, clothing, pose, or other attributes; normalize presentation conditions where feasible. Document the chosen photos and why the array was considered fair. The National Academies and Carlson and Carlson’s 2025 NIJ project discuss procedural fairness.
Fillers are not independent alternative suspects, and choosing a filler ordinarily should be recorded as a filler identification rather than quietly relabeled as “no identification.” If a witness chooses an innocent filler in a laboratory experiment, the researcher knows the response is incorrect. In real investigations, that choice warns against treating the test as a successful recognition of the suspect. The distinction between suspect selections, filler selections, and lineup rejections matters for transparent evidence interpretation.
3. Use a blind or effectively blinded administrator
A blind administrator does not know which lineup member is the suspect. A blinded protocol prevents the administrator from knowing which photo the witness is seeing at the moment of presentation, for example through appropriately designed computer or folder methods. These arrangements reduce the opportunity for inadvertent cues, changes in tone, gestures, pauses, or feedback. The National Research Council recommends blinded administration; the U.S. Department of Justice’s 2017 department-wide photo-array policy adopted blind or blinded methods subject to exceptional circumstances.
Blind administration protects both the witness and the investigation: it makes a suspect selection less vulnerable to criticism that the officer subtly guided the decision. Where a blind administrator is unavailable, the agency should follow a documented alternative that meaningfully shields suspect position and officer influence. Simply telling an administrator to remain neutral does not achieve the same protection against unconscious expectation effects.
4. Give neutral, standardized instructions before the test
Explain that the perpetrator may or may not be present, that the witness is free to reject the entire array, and that the investigation will continue whether or not anyone is chosen. Make clear that the administrator does not know the correct answer when the procedure is blind. Do not suggest that investigators have already captured the offender. Ensure the instructions are comprehensible to the witness, with interpreter and accessibility accommodations when needed. The National Academies recommendations call for consistent, easily understood instructions.
The option to say “I do not know” is a legitimate response, but its exact phrasing should not be presented as a universally proven fix. NIJ-sponsored research published in 2025 found that a specific “not sure” instruction did not improve eyewitness decision-making overall across its tests, despite some context-dependent protective effects. The most defensible approach is an evidence-informed standardized script rather than ad hoc pressure to pick or reject.
5. Record the first response and confidence in the witness’s words
The administrator should document precisely whom the witness selected, whether no person was selected, whether a filler was selected, the witness’s spontaneous explanation, and the witness’s confidence immediately after the decision, before confirmatory or corrective feedback. Capture the original wording rather than replacing it with an officer’s adjective such as “certain.” Preserve timing and circumstances and, when possible, video-record the entire session, including instructions and any remarks before and after identification. The National Academies and AP-LS policy statement specifically emphasize contemporaneous confidence and recording.
A later declaration of total certainty may carry less evidentiary information if the witness has seen the defendant in court, read reports, discussed the case, or received confirmation from investigators. Initial confidence is most informative when the first test was uncontaminated, fair, and blind; later courtroom confidence cannot be substituted automatically. This conditional interpretation is central to the Wixted–Wells scientific synthesis.
6. Avoid repeats, feedback, and contamination
Repeated exposure to the same suspect through mugshot books, successive lineups, video stills, social posts, or in-court confrontation can create familiarity with a previously seen image or face. Investigators should ordinarily avoid retesting the same witness with the same suspect, and should document any unavoidable exposures. Witnesses should not compare recollections with one another before independent statements and identifications have been secured. The 2020 consensus recommendations identify repeat identification attempts as a major avoidable source of distortion.
After the procedure, administrators should avoid statements such as “you picked our suspect,” “good job,” or “that matches the other witness.” Such feedback may alter the witness’s perceived certainty or retrospective estimate of how well the event was seen. Investigators can treat the result respectfully without validating a choice. Preserve the complete history of identification attempts, including prior rejections, changes of mind, and any eyewitness exposure to outside information.
Special Safeguards for Showups
When a showup is genuinely required by immediate operational circumstances, the administrator should document why a fair lineup was not feasible, how soon the showup occurred, what the witness had already been told, and what the suspect’s presentation looked like. The witness should receive a neutral warning that the person may or may not be the perpetrator, without statements asserting or implying that the person is guilty. Limit suggestive displays where safety and law permit, separate witnesses, collect the initial confidence statement verbatim, and record the entire process. These safeguards reflect the principles of the NIJ eyewitness guide and AP-LS guidance.
A showup rejection can carry investigative value, as can an identification, but both results require contextual evaluation. Comparing showups and lineups is more complicated than comparing raw selection percentages because choosing a filler is categorically different from accusing an innocent suspect. The research on showups versus lineups highlights how analytical choices change apparent error rates. A showup obtained under pressure or suggestive conditions warrants particular caution even when the witness sounds convincing.
What Determines Identification Accuracy?
Estimator variables: conditions investigators cannot retroactively change
Estimator variables arise from the witnessing event and the witness’s circumstances. They include viewing distance, duration, lighting, obstruction, disguise, divided attention, stress, delay, prior familiarity, and potentially cross-race face recognition. These affect the quality of information initially encoded or retained. Their presence changes the strength of the evidence but seldom supplies a reliable case-specific correction factor. The National Academies research synthesis distinguishes such event conditions from conditions created by investigators.
The cross-race effect refers to an average recognition disadvantage for unfamiliar faces from racial groups other than the observer’s own, with substantial variation across witnesses and experiences. It does not imply that any individual witness is necessarily wrong or that race determines credibility. A 2025 experimental comparison of lineups and showups observed a cross-race disadvantage in both formats; its findings describe group-level performance and should not be used to make categorical assumptions about a person.
System variables: conditions the justice system can control
System variables include the number and selection of fillers, instructions, administrator knowledge, presentation format, repeated testing, information leakage, and how confidence is elicited. Their central importance is practical: because they are controllable, policy can reduce avoidable bias even though no procedure recovers details that were never encoded. The AP-LS recommendations address nine policy areas, and the 2025 national agency survey shows that implementation remains uneven.
Confidence, accuracy, and the first uncontaminated test
Confidence is a witness’s subjective estimate of the identification decision. Under appropriately constructed and uncontaminated conditions, initial confidence can be statistically informative, and high-confidence identifications are often more accurate than low-confidence choices. This conclusion does not generalize automatically to unfair arrays, repeat tests, showups under suggestive circumstances, or testimony years later. Wixted and Wells (2017) describe the prerequisites and limits of confidence–accuracy calibration, while a 2025 preregistered experimental study found initial confidence more consistently informative than response speed.
Confidence–accuracy calibration is an empirical relationship across decisions and confidence bands in a defined sample. It is not a direct probability of guilt for a defendant. A high-confidence witness may still be mistaken, especially if the identification conditions were contaminated. Conversely, a low-confidence choice may convey uncertainty that investigators should preserve rather than suppress. A verdict requires lawful evaluation of all evidence, independent corroboration, and the governing standard of proof.
How Scientists Measure Identification Performance
The central outcome categories in controlled research are correct identification of a guilty suspect, false identification of an innocent suspect, selection of a filler, and rejection of the lineup. Researchers compare target-present with target-absent conditions where ground truth is known. Discriminability concerns how well a system separates guilty from innocent suspects; response bias concerns a witness’s willingness to select someone. Both contribute to raw choosing rates, so a procedure with fewer selections is not necessarily a procedure with better discrimination.
Receiver operating characteristic analysis evaluates correct and false suspect-identification rates across response thresholds. Confidence–accuracy characteristic analysis asks how often suspect selections made at different confidence levels are correct. The positive predictive value of an identification also depends on the underlying proportion of guilty suspects among the people tested, which is not ordinarily known with certainty in an actual investigation. Wixted and Mickes’ 2025 laboratory–field synthesis discusses these differences and the modeling needed when ground truth is unavailable.
An identification protocol’s reported performance from one laboratory population, police department, or experimental setup cannot simply be transferred into another jurisdiction as if error rates were constant. Changes in lineup composition, suspect-selection practices, witness demographics, case mix, familiarity, recording quality, and historical contamination can change accuracy. A field study often captures genuine investigative conditions but cannot directly label every suspect as guilty or innocent; a laboratory study knows ground truth but lacks some real stakes. Sound forensic reasoning uses both kinds of evidence without treating either as a universal case-specific diagnostic test.
What Current U.S. Practice Looks Like
The 2025 National Institute of Justice-supported survey provides an unusually useful reality check. Of 187 responding law-enforcement agencies in a survey conducted in 2020, 86% reported conducting photo arrays and 64% showups. Among agencies using photo arrays, 84% reported gathering witness confidence statements, compared with 72% among agencies using showups. For photo-array presentation, 61% reported sequential and 24% simultaneous presentation, while 13% allowed either. These figures describe responding agencies, and the 22.3% response rate limits confident generalization to every U.S. law-enforcement body.
The survey also found gaps in awareness of contemporary research, including the conditional usefulness of immediate confidence and the lack of a clear universal sequential advantage. Thus the existence of a published recommendation does not guarantee local implementation. For any case, the operative questions are which policy applied, whether it was followed, what was actually done, and what can be reconstructed from contemporaneous records.
Eyewitness Identification in Court: Science and Law
Psychological researchers assess memory and procedures; courts decide admissibility, constitutional safeguards, the allocation of proof, and the ultimate legal findings. In U.S. federal constitutional doctrine, Perry v. New Hampshire (2012) held that the Due Process Clause does not require the particular preliminary judicial reliability screen in the absence of unnecessarily suggestive circumstances arranged by law enforcement. This is a doctrine about constitutional screening, not a scientific finding that other suggestive influences leave eyewitness memory unaffected.
State rules may offer additional protections through statute, evidence law, police regulations, expert testimony, jury instructions, or decisions on identification procedures. For instance, the Massachusetts Supreme Judicial Court’s 2026 instructions address confidence, prior familiarity, exposure to outside information, and the form of identification procedure. Different jurisdictions may use different judicial tests and remedies. A claim that a particular lineup was “illegal” or that an identification must be excluded needs an applicable local legal basis.
Judges and jurors may benefit from a transparent chronology: the original witness description, the basis for choosing a suspect, the exact lineup images, the administrator’s knowledge, instructions, video or audio, the first decision, first confidence statement, later exposures, and changes across statements. An expert can explain established research and methodological limits, while the judge determines the legal admissibility of expert opinion and the fact-finder weighs evidence under the governing standard. The National Research Council recommends better training, pretrial inquiry, and educational tools for courts.
Common Errors and Misconceptions
“An honest eyewitness cannot be wrong.”
Honesty concerns intent; identification accuracy concerns memory performance under specific conditions. People can be completely sincere and mistaken. Procedures therefore should preserve the first uncontaminated response rather than equate apparent sincerity with reliability. The National Academies report explains the cognitive reasons eyewitness evidence requires careful collection.
“A confident eyewitness must be correct.”
A witness’s initial confidence can be informative in a fair, blind, first identification, and may lose interpretive value after suggestions, repetitions, feedback, or publicity. A confident identification from a compromised procedure warrants careful scrutiny. Confidence is evidence about a memory decision; it is never a finding of guilt. See the 2017 scientific synthesis.
“Sequential lineups are always scientifically superior.”
No universal ranking follows from the current literature. Investigators and courts should assess how the protocol was implemented and what specific outcomes the supporting research actually measured. The 2025 review describes both the method-dependent findings and uncertainty that remain.
“An immediate showup is automatically accurate.”
Speed may preserve temporal proximity, but showing only one person creates risks of suggestion, particularly if the witness is told police have detained the offender. A necessary showup should be independently justified, neutrally administered, documented, and interpreted in context. The AP-LS guidelines recommend minimizing unnecessary showups.
“No identification means the suspect is innocent or guilty.”
A lineup rejection, filler choice, or uncertain response is data about that testing event, not a legal finding. Investigators should preserve these outcomes because they may limit the evidentiary strength of a later identification. Whether a particular outcome bears on guilt depends on the complete evidentiary context and the reliability of the test.
A Practical Audit Checklist for Investigators, Attorneys, and Courts
A defensible file permits an independent reviewer to answer a sequence of questions. What independent facts linked the person to the investigation before identification? Was the witness’s initial description preserved before seeing photos or hearing suspect details? Did the array contain one suspect and adequately matched fillers? Who knew the suspect’s position? Exactly what instructions were given, and did the witness understand that the perpetrator might be absent? Was the full procedure recorded? What was the first selection, rejection, or uncertainty, and what were the witness’s own immediate words of confidence? Was there feedback or prior exposure to the suspect? Were other witnesses separated? Were later identifications or new information documented? A procedure should be judged from those contemporaneous facts, as recommended by the National Academies and AP-LS.
For showups, add the operational reason for proceeding without fillers, the time and location, how custody or restraints were presented, all relevant statements made to the witness, and whether the result was preserved in the first witness’s language. For complex cross-jurisdictional cases, determine which policies and evidentiary rules actually governed the procedure rather than importing another state’s protocol wholesale.
Ethical and Practical Implications
Eyewitness safeguards support accurate investigations, public trust, fair treatment of suspects, and appropriate respect for witnesses. Clear warnings and the ability to reject an array reduce pressure to guess. Blind administration protects against inadvertent guidance. Fair fillers make the test more informative. Full recordings allow investigators, prosecutors, defense counsel, courts, and experts to evaluate what actually happened rather than depend on retrospective summaries. These benefits follow from reducing controllable sources of evidence contamination; they do not imply that eyewitness evidence is uniformly unreliable or that witnesses should be disbelieved as a class.
These procedures also require accommodations. A witness who needs an interpreter, has a visual disability, is a child, or faces significant distress may require adapted communication and specialized expertise. Investigators should document the accommodation and any consequent departures from standard protocols. Research in one adult study population cannot be assumed to supply a validated accuracy estimate for children or for every cultural, linguistic, and clinical context. The goal is a procedure matched to the witness’s needs while preserving scientific transparency.
Frequently Asked Questions
What is the difference between a lineup and a showup?
A lineup offers a suspect among several fillers, giving the witness multiple plausible alternatives and an option to reject the entire set. A showup presents one person, usually in the field, without fillers. Because its single-person presentation is inherently suggestive, professional guidance recommends a fair lineup when feasible.
How many people should be in a photo lineup?
A common research-informed design uses one suspect and at least five suitably selected fillers, but exact procedural requirements vary. A visually balanced array matters more than treating the number alone as proof of fairness. Check the governing policy and preserve the complete array for review.
Why should a lineup administrator be blind?
An officer who knows the suspect’s identity or placement may inadvertently influence the witness by voice, posture, attention, or feedback. Blind or meaningfully blinded administration reduces those opportunities and strengthens documentation of an independent decision.
Can a witness identify the wrong person with high confidence?
Yes. High-confidence errors occur, and confidence is especially vulnerable to interpretation after suggestive procedures or repeated exposure. The relationship between initial confidence and accuracy is strongest under specified fair, uncontaminated conditions, and still describes statistical patterns rather than certainty in a particular case.
Should police use sequential or simultaneous lineups?
Current evidence does not establish a universally superior presentation method across all operational conditions. Agencies should follow a scientifically informed, standardized local policy and prioritize blind administration, fair construction, neutral instructions, initial confidence documentation, and complete recording.
Can a lineup rejection rule out a suspect?
A rejection is relevant evidence from a memory test and should be documented accurately. Its strength depends on viewing conditions, test fairness, time since the event, and other facts. Rejection does not by itself constitute a legal finding of innocence; it can properly affect the direction or weight of further investigation.
Are eyewitness identification rules the same in every U.S. state?
No. Constitutional doctrine supplies some common federal constraints, while states and agencies may provide additional requirements on showups, lineup procedures, evidence disclosure, admissibility, and jury instructions. The legal implications of a flawed procedure require jurisdiction-specific analysis.
What is the single most important record to preserve?
The strongest record is the complete first identification session, including the pretest description and instructions, full images or participants, administrator identity and knowledge, the actual response, the immediate confidence statement in the witness’s words, and an uninterrupted audiovisual record where practicable. Each component protects a different aspect of interpretability.
Conclusion
Eyewitness identification is most valuable when it functions as a carefully controlled recognition test: a genuinely supported suspect, a fair array where feasible, a blind administrator, neutral warnings, one initial decision, an immediate uncoached confidence statement, and complete preservation of the procedure. Showups require special justification and safeguards. Scientific results should be interpreted in the population and design that produced them, and legal consequences must be determined by the governing jurisdiction. The integrity of the process determines how much can responsibly be learned from the witness’s response.
For broader context, see Eyewitness Testimony: How Memory Works in Legal Cases.
Related Articles
References
Amendola, K. L., Olson, M., Carlson, C., Gronlund, S., Mickes, L., Gao, J., & Yang, Y. (2025). Nationwide Survey of Policies and Practices in Eyewitness Identification: Full Report. National Institute of Justice. Official report.
Carlson, C. A., & Carlson, M. A. (2025). Eyewitness Identification: A Systematic Investigation of Lineup Composition and Fairness. National Institute of Justice. Official report.
Charman, S. (2025). Testing a “Not Sure” Instruction to Reduce the Harmful Impact of System and Estimator Variables on Lineup Identification Accuracy. National Institute of Justice. Official report.
Massachusetts Supreme Judicial Court. (2026). Statement of the Supreme Judicial Court on the Model Jury Instructions on Eyewitness Identification. Official judicial source.
National Research Council. (2014). Identifying the Culprit: Assessing Eyewitness Identification. National Academies Press. https://doi.org/10.17226/18891.
Perry v. New Hampshire, 565 U.S. 228 (2012). Supreme Court opinion.
Shen, K. J. (2025). The Cross-Race Effect in Lineups Versus Showups. Applied Cognitive Psychology. https://doi.org/10.1002/acp.70035.
Technical Working Group for Eyewitness Evidence. (1999). Eyewitness Evidence: A Guide for Law Enforcement. National Institute of Justice. Official handbook.
U.S. Department of Justice. (2017). Justice Department Announces Department-Wide Procedures for Eyewitness Identification. Official release.
Wells, G. L., Kovera, M. B., Douglass, A. B., Brewer, N., Meissner, C. A., & Wixted, J. T. (2020). Policy and Procedure Recommendations for the Collection and Preservation of Eyewitness Identification Evidence. Law and Human Behavior, 44(1), 3–36. https://doi.org/10.1037/lhb0000359.
Wixted, J. T., & Mickes, L. (2025). Suspect identification accuracy from lineups, in the lab and in the field. Cognitive Research: Principles and Implications, 10, 60. https://doi.org/10.1186/s41235-025-00670-1.
Wixted, J. T., & Wells, G. L. (2017). The Relationship Between Eyewitness Confidence and Identification Accuracy: A New Synthesis. Psychological Science in the Public Interest, 18(1), 10–65. https://doi.org/10.1177/1529100616686966.