BACKGROUND: Poor adherence to medication limits the effectiveness of treatment for human immunodeficiency virus. Systematic reviews can identify practical and effective interventions. Meta-analyses that control for variability in standard care provided to control groups may produce more accurate estimates of intervention effects. METHODS: To examine whether viral load and adherence success rates could be accurately explained by the active content of highly active antiretroviral therapy (HAART) adherence interventions when controlling for variability in care delivered to controls, databases were searched for randomized controlled trials of HAART adherence interventions published from 1996 to January 2009. A total of 1342 records were retrieved, and 52 articles were examined in detail. Directly observed therapy and interventions targeting specific patient groups (ie, psychiatric or addicted patients, patients <18 years) were excluded, yielding a final sample of 31 trials. Two coders independently retrieved study details. Authors were contacted to complete missing data. RESULTS: Twenty studies were included in the analyses. The content of adherence care provided to control and intervention groups predicted viral load and adherence success rates in both conditions (P < .001 for all comparisons), with an estimated impact of optimal adherence care of 55 percentage points. After controlling for variability in care provided to controls, the capacity of the interventions accurately predicted viral load and adherence effect sizes (R(2) = 0.78, P = .02; R(2) = 0.28, P < .01). Although interventions were generally beneficial, their effectiveness reduced noticeably with increasing levels of standard care. CONCLUSIONS: Intervention and control patients were exposed to effective adherence care. Future meta-analyses of (behavior change) interventions should control for variability in care delivered to active controls. Clinical practice may be best served by implementing current best practice.