Early prediction of suboptimal viral response to highly active antiretroviral therapy (HAART) is vital to prevent early development of drug resistance. We used logistic regression to predict the odds of achieving virologic suppression (<50 copies/mL) after 24 weeks of HAART in 656 antiretroviral-naive patients starting HAART at the J.W. Goethe University, Chelsea and Westminster, and Royal Free Hospitals according to their week 4 viral load. Therapy changes involving the switch of a single antiretroviral were assumed to have occurred for toxicity reasons and ignored. Because complete regimen changes or additions of new antiretrovirals could be due to virologic failure, patients were counted as virological failures at week 24. Three hundred sixty (84%) of 430 patients with viral loads of <1000 copies/mL, 106 (61%) of 175 with viral loads between 1001 and 10,000 copies/mL, 11 (37%) of 30 with viral loads between 10,001 and 100,000 copies/mL, and 5 (24%) of 21 with viral loads of >100,000 copies/mL at week 4 subsequently attained virologic suppression at 24 weeks. The odds of attaining virologic suppression at 24 weeks was 65% lower for every 1-log higher viral load at week 4 (odds ratio, 0.35; 95% confidence interval, 0.27-0.45). The proportion of patients with an undetectable viral load at 24 weeks among those who have not attained a viral load of <1000 copies/mL by 4 weeks is quite low. We suggest that this group of patients should be particularly closely monitored.