Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
The Rey Complex Figure Test (RCFT) is a neuropsychological drawing task in which a person copies a fixed, abstract line figure and then redraws it from memory. Researchers use it as a measure of visuospatial constructional ability and visuospatial memory. This guide is for researchers, students and study coordinators who need to know how it is administered, how the 36-point score is built, how the common scoring systems differ, and how to report results. It is general reference information, not a way to interpret anyone’s personal results or to diagnose a condition.
What the Rey Complex Figure Test is
André Rey developed the figure in 1941, and Paul-Alexandre Osterrieth standardized its administration and scoring in 1944 (preliminary data from 230 children and 60 adults) (Li, 2021). The task is sometimes called the Rey-Osterrieth Complex Figure (ROCF) or RCFT. Copying the figure draws on attention, fine-motor coordination, visuospatial perception, non-verbal memory, planning and organization, and spatial orientation (Li, 2021), which is why a single score cannot be read as a measure of one ability.
The commercial version most often cited in current research is the Rey Complex Figure Test and Recognition Trial (RCFT) by John E. Meyers and Kelly R. Meyers, published by PAR. PAR describes it as measuring visuospatial ability and visuospatial memory, for ages 6 to 89, in about 45 minutes including a timed 30-minute delay (PAR, accessed 2026-11-07).
Trials and administration
The PAR edition has four trials: Copy, a 3-minute Immediate Recall, a 30-minute Delayed Recall, and a Recognition trial. A stopwatch is required, the standardized stimulus card is 8.5 x 11 inches (the norms were collected with that card), and the test is available in print and as an in-person tablet kit. PAR assigns it Qualification Level C (PAR, accessed 2026-11-07).
| Trial | What the examinee does |
|---|---|
| Copy | Copies the figure |
| Immediate Recall | Redraws from memory after 3 minutes |
| Delayed Recall | Redraws from memory after 30 minutes |
| Recognition | Recognition of the figure’s elements (the Meyers trial uses 24 patterns) |
PAR lists the measured domains as visuospatial recall memory, recognition memory, response bias, processing speed and visuospatial constructional ability.
Scoring formula: 36 points from 18 units
In the Osterrieth system the figure is divided into 18 scoring units, each worth a maximum of 2 points, for a maximum of 36. Partial credit is given when a unit is accurate but misplaced, or recognizable but inaccurate (Li, 2021). In the PAR edition each unit is rated for accuracy and placement, and the same criteria apply to all three drawing trials (PAR, accessed 2026-11-07). The unit definitions and exact credit rules are in the licensed manual, which this page does not reproduce.
Worked example of the arithmetic (illustrative values)
The unit scores below are invented to show the arithmetic only. They are not data from any person or study. Suppose a drawing is rated as follows: 12 units at 2 points, 4 units at 1 point, and 2 units at 0 points.
- 12 x 2 = 24
- 4 x 1 = 4
- 2 x 0 = 0
- Total = 24 + 4 + 0 = 28 of 36, or 28 / 36 = 77.8% of the maximum.
The raw total says how many points were earned, not how typical that is. It becomes interpretable only against norms for the person’s age and, where the norms allow, education.
Scoring systems compared
| System | What it adds or emphasizes |
|---|---|
| Osterrieth | 18 units, 0 to 2 points each, 36 maximum; partial credit rules |
| Taylor | More detailed rules for accuracy and location |
| Meyers | Clearer accuracy and location criteria plus a 24-pattern recognition test |
| Boston Qualitative Scoring System (BQSS) | Process-based: fragmentation, planning, neatness, perseveration |
Reported inter-rater reliability is 99% for the Osterrieth system and 94% for the Meyers system (Li, 2021). For the PAR edition, PAR reports test-retest reliability of .76 to .89 for memory scores and interrater reliability of .93 to .99 (PAR, accessed 2026-11-07). A simplified Taylor figure has reported internal consistency of .89 (copy), .97 (immediate recall) and .96 (delayed recall) (Li, 2021). Accuracy-based totals can miss organizational strategy and executive function, while process scoring captures them but is time-consuming (Li, 2021). For background on these reliability terms, see our guides on reliability in research measurement and reliability vs validity.
Interpretation, norms and results
The PAR edition provides demographically corrected norms for the copy and memory variables (PAR, accessed 2026-11-07). That matters because education strongly affects performance and scores are lower in older adults, so the literature recommends education-stratified cutoffs (Li, 2021). A single raw cutoff applied to every sample is therefore a design weakness.
| Result | What it can support | What it cannot show |
|---|---|---|
| Low copy score | Difficulty with constructional or visuospatial organization on that occasion | Which underlying process failed, because several processes feed the copy |
| Total of 36 | Maximum accuracy under the scoring rules | Planning strategy, which needs process-based scoring |
An RCFT score is one data point to combine with other measures, and a raw total is not a diagnosis. Two related guides help with the statistical side of using such a score in a study: floor and ceiling effects (relevant when a sample clusters near 0 or near 36) and convergent and discriminant validity.
Assumptions and common mistakes
- Reading a raw total as if it were normed. A raw score of 28 of 36 says how many points were earned, not how typical that is. It needs comparison with norms for age and, where available, education.
- Using one cutoff for every education level. Education strongly affects performance and scores are lower in older adults, so the literature recommends education-stratified cutoffs (Li, 2021).
- Treating a printout or on-screen copy as equivalent to the standardized card. The PAR norms were collected with the standardized 8.5 x 11 inch stimulus card, so other formats are not equivalent (PAR, accessed 2026-11-07).
- Mixing scoring systems. Osterrieth, Taylor, Meyers and BQSS use different rules, so scores and reliability figures from one system should not be pooled or compared with another without saying so. Report which system you used.
- Rater drift. Raters can apply the criteria differently over time or from one another. Train raters on the manual, check agreement on a subset of drawings, and report the agreement value and how disagreements were resolved.
- Treating one score as a diagnosis. An RCFT score is one data point to combine with other measures.
How to report RCFT results (APA style)
State the edition and scoring system, the trials administered, the examiner qualification, the norms used, and whether you analyze raw scores or norm-referenced scores. A template, with placeholders in brackets:
Visuospatial memory was assessed with the Rey Complex Figure Test and Recognition Trial (Meyers and Meyers, [year of your manual]), scored with the 36-point system by [number] trained raters; inter-rater agreement was [value]. Analyses used [raw / demographically corrected] scores for the Copy, Immediate Recall and Delayed Recall trials.
Report how disagreements between raters were resolved, and cite the manual edition you used.
Adding up the 18 unit scores in Excel, Python or R
Excel: with the 18 unit scores in cells A1:A18, use =SUM(A1:A18). Python, with the same illustrative values as above:
units = [2]*12 + [1]*4 + [0]*2
assert len(units) == 18
total = sum(units) # 28
pct = total / 36 * 100 # 77.8
R: units = c(rep(2,12), rep(1,4), rep(0,2)); sum(units) returns 28.
Access, licensing, and PDF or online versions
The PAR edition is a Qualification Level C product, restricted to appropriately qualified purchasers (PAR, accessed 2026-11-07). It is administered in person with a stopwatch, and the norms were collected with the standardized stimulus card, so a printout or an on-screen copy of the figure is not equivalent. The figure, record forms and scoring manual are commercial publications; researchers who need the instrument should obtain it, and any required training, from the publisher.
Frequently asked questions
What does the Rey Complex Figure Test measure?
Visuospatial constructional ability and visuospatial memory, with the PAR edition also reporting recognition memory, response bias and processing speed (PAR, accessed 2026-11-07).
How is the Rey Complex Figure Test scored?
By rating 18 units of the drawing for accuracy and placement, up to 2 points each, for a 36-point maximum per drawing trial (Li, 2021; PAR, accessed 2026-11-07).
Can I interpret my own RCFT result?
No. A score depends on age and education norms and on clinical context, and it has to be read by a qualified professional. This page is general reference information.
Is there an RCFT PDF or a free online test?
The materials are published commercially and the test is administered in person by a qualified examiner; see the access section above.
Sources
- PAR, Inc. Rey Complex Figure Test and Recognition Trial (RCFT) product page, parinc.com/RCFT, accessed 2026-11-07.
- Li, Y. (2021). Overview of the Complex Figure Test and its clinical application in neuropsychiatric disorders, including copying and recall. Frontiers in Neurology, 12, 680474. doi.org/10.3389/fneur.2021.680474, accessed 2026-11-07.







