This study investigated the effects of two types of control over instruction (learner and program) and two modes of instructional programs (lean and full) on the achievement, option use, and time-in-program of 274 high-ability and low-ability students from grades 9 and 10. The basic instructional program in geometry was delivered by computer. Subjects under learner control scored significantly higher on the program posttest than those under program control, spent significantly more time in the program, and liked it better. Learner-control subjects appeared to “trust” their given version of the program, viewing many more optional screens in the full version than those in the lean one. High-ability learners adapted their study behavior to the lean version under learner control by choosing significantly more optional screens than their low-ability counterparts.