OBJECTIVE Reproducibility of results achieved by psychological performance tests is a precondition to quantify reliable cognitive and psychomotoric behaviour alterations after interventions. However, evidence exists that repeated processing leads to training effects which influence results. Therefore it is essential to elucidate these training effects. METHODS Ten subjects (6 men and 4 women, 42 +/- 12, 5 years) carried out two computer based performance tests of the "Wiener Testsystem" (WTS), the Determinationstest (DT) and the Arbeitsleistungsserie (ALS). Both tests were performed five times by each test person. Training effects and contrasts were analyzed with one-way analysis of variance (ANOVA) for repeated measures. RESULTS A significant "training" effect (p <or= 0.05%) (base-line/first repetition) can be demonstrated in both tests. Improvements of reactive force, capacity and alertness (DT Subtest 1 and Subtest 2) are observed even up to the fifth processing. Except in "reaction time" (Subtest 2) subjects significantly improved. Similarly improvements were found for concentration (ALS) at the first and third time of test handling. Failures in processing "Arbeitsleistungsserie" (ALS) decreased with the third but increased at fourth test handling compared to previous level. Considering mean values highest improvements were found from base-line to first repetition (7,8-21%). Thereafter improvements of about 6% were observed. CONCLUSIONS Repeated testing of computer-assisted cognitive and psychomotoric performance tests resulted in significant test improvements which are most likely caused by training effects and increased familiarity with the test system and therefore probably cannot be interpreted as real cognitive improvements. In conclusion, positive effects evaluated with performance tests, e.g. before and after interventions, should be interpreted with regard to these results.