Objectives: To determine the efficacy of adding abacavir (Ziagen, ABC) to optimal stable background antiretroviral therapy (SBG) to AIDS dementia complex (ADC) patients and address trial design. Design: Phase III randomized, double-blind placebo-controlled trial. Setting: Tertiary outpatient clinics. Participants: ADC patients on SBG for ≥8 wk. Interventions: Participants were randomized to ABC or matched placebo for 12 wk. Outcome Measures: The primary outcome measure was the change in the summary neuropsychological Z score (NPZ). Secondary measures were HIV RNA and the immune activation markers β-2 microglobulin, soluble tumor necrosis factor (TNF) receptor 2, and quinolinic acid. Results: 105 participants were enrolled. The median change in NPZ at week 12 was +0.76 for the ABC + SBG and +0.63 for the SBG groups (p = 0.735). The lack of efficacy was unlikely related to possible limited antiviral efficacy of ABC: at week 12 more ABC than placebo participants had plasma HIV RNA ≤400 copies/mL (p = 0.002). There were, however, other factors. Two thirds of patients were subsequently found to have had baseline resistance to ABC. Second, there was an unanticipated beneficial effect of SBG that extended beyond 8 wk to 5 mo, thereby rendering some of the patients at baseline unstable. Third, there was an unexpectedly large variability in neuropsychological performance that underpowered the study. Fourth, there was a relative lack of activity of ADC: 56% of all patients had baseline cerebrospinal fluid (CSF) HIV-1 RNA <100 copies/mL and 83% had CSF β-2 microglobulin <3 nmol/ L at baseline. Conclusions: The addition of ABC to SBG for ADC patients was not efficacious, possibly because of the inefficacy of ABC per se, baseline drug resistance, prolonged benefit from existing therapy, difficulties with sample size calculations, and lack of disease activity. Assessment of these trial design factors is critical in the design of future ADC trials.