The use of molecular genotyping to distinguish recrudescence from new infections has become common in antimalarial clinical trials. However, methods used to interpret genotyping results have not been standardized. We analyzed data from 3,000 patients enrolled in clinical trials at seven sites in Uganda. Late treatment failure requiring genotyping occurred in 51% of the patients. Among samples successfully genotyped, 21% were definitive new infections (no recrudescent strains present on day of failure), 35% were definitive recrudescences (only recrudescent strains present), and 44% were mixed (new and recrudescent strains present). The probability of having a mixed genotyping result increased as transmission intensity increased. At the highest transmission site, the estimated risk of treatment failure increased from 34% to 84% for chloroquine plus sulfadoxine-pyrimethamine, from 18% to 45% for amodiaquine plus sulfadoxine-pyrimethamine, and from 12% to 57% for amodiaquine plus artesunate, depending on whether mixed genotyping results were classified as new infections or recrudescences, respectively. The method used to classify treatment outcomes can have a major impact on estimates of drug efficacy, especially in areas of high transmission intensity.