BACKGROUND: We examined diaphragm adherence among 2429 women randomized to the intervention arm (diaphragm + gel + condoms) in Methods for Improving Reproductive Health in Africa, a phase III trial of the diaphragm for HIV prevention in Zimbabwe and South Africa. METHODS: Women were followed for a median of 7 quarterly visits (range: 1-8 quarterly visits) during which diaphragm adherence was assessed. We conducted trajectory analyses to identify behavioral groups associated with specific diaphragm adherence patterns. Multivariate multinomial logistic regression was used to identify baseline characteristics associated with higher probability of being in a particular trajectory group. RESULTS: Diaphragm uptake was very high (3.1% never used diaphragms). However, diaphragm adherence was reported at only 49% of visits. Women were clustered into 4 diaphragm adherence groups based on their highest estimated group membership probability: low adherers (31.0%), decreasing adherers (28.9%), increasing adherers (9.3%), and high adherers (30.8%). Women classified as high adherers (as compared with low adherers) were more likely to be older [adjusted odds ratio (AOR) = 1.09, 95% confidence interval (CI): 1.07 to 1.11] and to report baseline condom adherence (AOR = 2.00, 95% CI: 1.47 to 2.71). They were less likely to have high-risk behavior (AOR = 0.51; 95% CI: 0.37 to 0.71) and to have high-risk partners (AOR = 0.58; 95% CI: 0.43 to 0.78). They were most likely to be from the Zimbabwe site (AOR = 2.82; 95% CI: 1.89 to 4.20) and least likely to be from the Johannesburg site (AOR = 0.51; 95% CI: 0.37 to 0.77). CONCLUSION: This analytic approach could help to identify high compliers for enrollment in future HIV prevention trials or the types of participants who may need intensive adherence counseling during follow-up.