Using multi-process framework by McDaniel and Einstein (2000), the current study examined whether the length of prospective memory (PM) delay intervals as measured by the 2- and 15-min subscales of the Memory for Intentions Screening Test (MIST) have differential predictive value for antiretroviral (ARV) adherence. Participants included 74 HIV-infected individuals whose ARV adherence was tracked with an electronic monitoring system. Participants were classified as "adherent" (n = 49) or "non-adherent" (n = 25) based on recorded pill bottle openings of ≥90% of prescribed doses over 30 days. An adherence group by MIST delay interval interaction was observed, such that non-adherent participants had worse performance on the 15-min, but not 2-min delay PM MIST subscales. The observed MIST 15-min delay effects were significantly more pronounced on time- versus event-cued PM trials. Long-delay time-based PM was predictive of non-adherence independent of demographics, mood state, self-reported adherence, and general cognitive functioning. Findings from this clinical study indicate that ARV non-adherence may be particularly associated with deficits in strategic cue monitoring over longer PM delays, which may inform interventions to improve adherence among persons living with HIV infection.