Molecular genotyping of baseline and post-treatment recurrent Plasmodium falciparum is recommended to distinguish recrudescent from new infections. However, genotyping performance and adjustment of treatment outcomes have not been evaluated in large field trials. Parasitological outcomes were assessed in nine double-blinded trials of uncomplicated P. falciparum malaria in African children treated with artesunate/placebo plus standard monotherapies. Day 28 failure rates were adjusted by stepwise genotyping the P. falciparum glutamate rich protein (glurp), merozoite surface protein 1 (msp1) and 2 (msp2). We calculated overall and laboratory genotyping performance and compared unadjusted (crude) and PCR-adjusted outcomes. 3455 (93.6%) of 3691 enrolled patients were evaluable by Day 28. 767 (22%) had post-Day 14 recurrent parasitemias of which 686 could be genotyped: 246 were recrudescences, 286 new infections and 154 unresolved. The overall and laboratory genotyping performance were 69 (12-100)% and 78 (50-100)%, respectively. The mean Day 28 crude parasitological failure rate was 44 (range 3-87)%. PCR-adjusted rates were 36 (range 2-86)% if unresolved infections were counted as failures or 33 (range 2-86)% if excluded from analysis. The overall difference between crude Day 28 and Day 14 failure rates was 22% (95% CI 20.3, 24.6) but decreased to 14% (95% CI 12.1, 16.3) if unresolved infections are counted as failures, or to 11% (95% CI 9.8, 16.3) if unresolved infections are excluded from the analysis. Genotyping refined treatment outcomes but diligence is needed in sample collection and analysis to improve its performance. Our findings support the WHO recommendation of PCR genotyping in malaria clinical trials and suggest that stepwise genotyping of only two loci (msp2 and msp1 or glurp) can reliably discriminate recrudescences from new infections.