This study examined the effects of a prototype of motivationally-adaptive CAI (Computer Assisted Instruction), developed in accordance with the ARCS (Attention, Relevance, Confidence, Satisfaction) model of motivational design, on achievement, perceived motivation (both overall motivation and for each of the ARCS components), efficiency, and continuing motivation. Subjects were 66 tenth grade students. Students used one of three versions of CAI presenting material on genetics--motivationally-adaptive CAI, motivationally-saturated CAI, or motivationally-minimized CAI. Motivationally-adaptive CAI showed higher effectiveness, overall motivation, and attention than the other two CAIs. For relevance, motivationally-adaptive CAI was higher than motivationally-saturated CAI, but not higher than motivationally-minimized CAI. For confidence and satisfaction, motivationally-adaptive CAI was not higher than the other two CAIs. For efficiency, both motivationally-adaptive CAI and motivationally-minimized CAI were higher than motivationally-saturated CAI, but only the efficiency of the motivationally-adaptive CAI was identified as having practical importance. For continuing motivation, there were no significant differences among the three CAIs, but a significant correlation was found between overall motivation and continuing motivation across three CAIs. These findings support the contention that the ARCS model is useful for developing motivationally-adaptive CAI. Motivationally-adaptive CAI can be an effective, efficient, and motivating form of instruction that may also enhance learners' continuing motivation. (Contains 51 references.) (Author/MES)