Abstract Background This study aimed to explore Plasmodium falciparum population dynamics during the early phase of anti-malarial drug treatment with artemisinin-based combination therapy in children with clinical malaria in a high transmission area in Africa. Methods A total of 50 children aged 1-10 years with acute uncomplicated P. falciparum malaria in Bagamoyo District, Tanzania, were enrolled. Participants were hospitalized and received supervised standard treatment with artemether-lumefantrine according to body weight in six doses over 3 days. Blood samples were collected 11 times, i.e. at time of diagnosis (-2 h) and 0, 2, 4, 8, 16, 24, 36, 48, 60 and 72 h after initiation of treatment. Parasite population dynamics were assessed using nested polymerase chain reaction (PCR)-genotyping of merozoite surface protein (msp) 1 and 2. Results PCR-analyses from nine sequential blood samples collected after initiation of treatment identified 20 and 21 additional genotypes in 15/50 (30%) and 14/50 (28%) children with msp1 and msp2, respectively, non-detectable in the pre-treatment samples (-2 and 0 h combined). Some 15/20 (75%) and 14/21 (67%) of these genotypes were identified within 24 h, whereas 17/20 (85%) and 19/21 (90%) within 48 h for msp1 and msp2, respectively. The genotype profile was diverse, and varied considerably over time both within and between patients, molecular markers and their respective families. Conclusion PCR analyses from multiple blood samples collected during the early treatment phase revealed a complex picture of parasite sub-populations. This underlines the importance of interpreting PCR-outcomes with caution and suggests that the present use of PCR-adjustment from paired blood samples in anti-malarial drug trials may overestimate assessment of drug efficacy in high transmission areas in Africa. The study is registered at http://www.clinicaltrials.gov with identifier NCT00336375.