We studied three antimalarial treatments in Caala and Kuito, Angola, in 2002 and 2003. We tested chloroquine (CQ), amodiaquine (AQ) and sulfadoxine-pyrimethamine (SP) in Caala, and AQ, SP and the combinations AQ + artesunate (AQ + AS) and SP + artesunate (SP + AS) in Kuito. A total of 619 children (240 in Caala, 379 in Kuito) with uncomplicated Plasmodium falciparum malaria were followed-up for 28 days, with PCR genotyping to distinguish recrudescence from reinfection. PCR-corrected failure proportions at day 28 were very high in the CQ group (83.5%, 95% CI 74.1-90.5), high in the SP groups (Caala: 25.3%, 95% CI 16.7-35.8; Kuito: 38.8%, 95% CI 28.4-50.0), around 20% in the AQ groups (Caala: 17.3%, 95% CI 10.0-27.2; Kuito: 21.6%, 95% CI 14.3-30.6) and very low in the artemisinin-based combination groups (1.2%, 95% CI 0.0-6.4 for each combination AQ + AS and SP + AS). These results show that CQ and SP are no longer efficacious in Caala and Kuito and that the moderate efficacy of AQ is likely to be compromised in the short term if used as monotherapy. We recommend the use of AQ with AS, though this combination might not have a long useful therapeutic life because of AQ resistance. © 2005 Royal Society of Tropical Medicine and hygiene. Published by Elsevier Ltd. All rights reserved.