Adherence undeniably impacts product effectiveness in microbicide trials, but the connection has proven challenging to quantify using routinely collected behavioral data. We explored this relationship using a nested case–control study in the CAPRISA 004 Tenofovir (TFV) gel HIV prevention trial. Detailed 3-month recall data on sex events, condom and gel use were collected from 72 incident cases and 205 uninfected controls. We then assessed how the relationship between self-reported adherence and HIV acquisition differed between the TFV and placebo gel groups, an interaction effect that should exist if effectiveness increases with adherence. The CAPRISA 004 trial determined that randomization to TFV gel was associated with a significant reduction in risk of HIV acquisition. In our nested case–control study, however, we did not observe a meaningful decrease in the relative odds of infection—TFV versus placebo—as self-reported adherence increased. To the contrary, exploratory sub-group analysis of the case–control data identified greater evidence for a protective effect of TFV gel among participants reporting less than 80 % adherence to the protocol-defined regimen (odds ratio (OR) 0.30; 95 % CI 0.11–0.78) than among those reporting ≥80 % adherence (Odds Ratio 0.81; 95 % CI 0.34–1.92). The small number of cases may have inhibited our ability to detect the hypothesized interaction between adherence and effectiveness. Nonetheless, our results re-emphasize the challenges faced by investigators when adherence may be miss-measured, miss-reported, or confounded with the risk of HIV.