Two versions of a computer-assisted instruction practice lesson on series resonant circuit analysis were compared. One was a program control version which required considerable time for students to reach criterion level. The other program was a revised student control version of the same lesson. In the program control version, quantity of practice was determined by an adaptive branching algorithm based on student performance; the student control version made practice prior to a criterion item optional. Subjects were 84 trainees in basic electricity/electronics at the Naval Training Center in San Diego. Those using the student control program showed a marked time savings over those using program control, and student control of training produced more uniform performance. Large time savings and increased test scores supported the hypothesis that a knowledgeable student can realistically estimate when he has mastered the practice material. (CH)