OBJECTIVE: The complexity of highly active antiretroviral therapy (HAART), with multiple medications, formulations, and dosing intervals, makes adherence challenging. Little is known about the adherence of children to HAART. The objective of this study was to identify correlates of adherence to HAART and the relationship between adherence and study outcomes in a pediatric clinical trial. METHODS: Pediatric AIDS Clinical Trials Group 377 is a phase I/II randomized trial of 4 HAART regimens in antiretroviral-experienced, clinically stable children aged 4 months to 17 years. The 4 treatment arms include various 3- or 4-drug combinations of d4T, 3TC, nevirapine, ritonavir, and nelfinavir. After informed consent was obtained, 193 children were enrolled between December 1997 and September 1998. Questionnaires were developed to collect subject- or caregiver-reported adherence to study medications and to identify problems associated with medication administration. Every 3 months, the number of doses of each medication missed during the 3 days preceding the study visit was recorded. Full adherence (FA) and non-full adherence were defined as missing no doses and missing at least 1 dose, respectively. RESULTS: Adherence data from study week 48 or the most recent study visit were available for 125 children (week 48 for 109 children). Overall, 70% of children reported FA and 30% reported non-full adherence. Adherence did not differ by treatment arm, age, or the child's knowledge of his or her human immunodeficiency virus infection status. There was a suggestion that adherence was less for white than nonwhite children (40% vs 73% FA) and did not differ between black and Hispanic children. Rates of FA were 82% for d4T, 79% for 3TC, 83% for nevirapine, 84% for ritonavir, and 68% for nelfinavir. Despite the similar rates of FA, difficulties with taking specific medications were reported most frequently for ritonavir and nelfinavir. These included poor taste, patient refusal, and scheduling problems. Adherence was associated with the virologic response: FA was seen in 92% of children with > or =2 log10 drop in viral load and in 64% with <2 log10 drop in viral load. CONCLUSION: In children, reported adherence predicts the virologic response to HAART therapy and is a useful measure of adherence. Interventions and regimens to increase adherence to HAART should result in an improved outcome.