This study evaluated the cure rate of the standard recommended regimen for Plasmodium vivax malaria in Brazil and assessed risk factors for failures. Fifty patients with vivax malaria given supervised medical treatment (standard dose of chloroquine: total dose = 1.5 g over a three-day period plus primaquine: total dose = 210 mg over a 14-day period) were followed for six months in a non-endemic area. Cox's regression was used to identify predictors of relapses. Among the 289 patient-months of follow-up, seven relapses were identified (2.4 relapses per 100 person-months) between 33 and 137 days after treatment initiation. Risk factors for relapses (P < or = 0.05) were female sex, higher parasitemia at baseline, shorter number of days with symptoms prior to baseline, and lower mg/kg dose of primaquine. Relapses following supervised vivax treatment is in principle a necessary, but not sufficient, component of in vivo parasite resistance. Results indicate that other factors, principally sub-therapeutic primaquine doses, may explain the occurrence of vivax treatment failures.