A study was conducted (1) to obtain an estimate of the amount of error in prediction that occurs when static measures are used to interpret work-sample performance; (2) to identify a number of dynamic performance measures that reflect learning that might occur on the task; and (3) to compare the predictive accuracy of those measures to that of static measures. The participants in the study were 10 male and 10 female handicapped vocational evaluation clients. The clients practiced on a work sample involving a relatively simple psychomotor task for 5 consecutive work days (50 trials/day). The data collected showed that the participants improved an average of 31 percent in performance speed over the 5 days, with 11 exceeding the industrial standard by day 5. The accuracy of eight methods of predicting day 5 performance was investigated; all of the dynamic prediction methods proved superior to the traditional static work-sample measure. It was concluded that the use of day 1 total response-time scores seriously underestimates the level of performance that a handicapped individual can potentially achieve on a task following practice, while analyzing the performance data with dynamic performance measures can result in significantly more accurate estimates of the level of performance that someone can achieve. It was suggested that microcomputers be used in data collection and analysis of dynamic performance measures. (KC)