Compliance rates from four iterations of Student Bodies, an 8–10 week computer assisted health education (CAHE) program for the prevention of eating disorders in college-aged women, were measured. Each iteration was modified with the hope of increasing participant compliance. Participants were 116 undergraduate women from two large West Coast universities who were randomly assigned to either the CAHE program or a wait-list group. Model 1 was delivered via CD-ROM, was anonymous, used an unstructured “grazing” approach, and participants received $10. Model 2 was web-based, had recommended assignments with telephone reminders, and participants received $25. Model 3 used a more structured approach to assignments. Model 4 eliminated anonymity, added an academic component, used email reminders, and offered course units as an incentive. Compliance was defined as percentage of screens read per number of screens assigned (n = 27–33). In Model 1, compliance was self-reported at 53%. In Models 2, 3, and 4, compliance, measured directly via computer logs, was 66%, 85%, and 84%, respectively. A Mann-Whitney test showed that the increase in compliance between Models 2 and 3 and Models 2 and 4 was statistically significant (p < 0.01). Compliance was significantly correlated with improvement on outcome. Suggestions for increasing compliance in other CAHE programs are offered.