In an effort to compare different methods of instructing patients, 99 women 18-25 years of age were given computer, spoken, weitten, or no instructions for the collection of a clean voided urine specimen. The group who received computer instructions was the most uniform in its performance (P less than 0.002, F-test) and reported the fewest procedural problems (P less than 0.02, Fisher test). In addition, this group had fewer contaminating bacteria than the group who received written instructions (P less than 0.03 , Mann-Whitney test). The group who received no instructions had more bacteria (P less than 0.0001, Mann-Whitney test) than any of the other groups. The effectiveness of the computer instruction was probably related to numerous attributes, including the individualized quality of the dialogue, self-pacing, self-testing, and privacy.