BACKGROUND: Parasite resistance to the anti-malarial drug chloroquine is common in eastern Sudan. Dynamic within-host changes in the relative abundance of both sensitive and resistant Plasmodium falciparum parasites were examined in a cohort of chloroquine-treated patients presenting with uncomplicated falciparum malaria, using a novel allele-specific quantitative approach. METHODS: Treatment outcomes were determined for 93 patients of all ages in a per protocol cohort using a modified 14-day WHO protocol. Parasite DNA samples at days 0, 1, 2, 3, 7 and 14 following treatment were analysed using real-time quantitative PCR methods that distinguished resistant and sensitive genotypes at amino acids 72-76 of the pfcrt locus. RESULTS: Chloroquine treatment was not efficacious, and of 93 assessable patients, only 10 individuals (10.7%; 95% C.I. 4.34-17.2%) enjoyed an adequate clinical and parasitological response. Resistant parasites with the haplotype CVIET at codons 72-76 of the pfcrt locus were dominant in the starting population. Chloroquine sensitive parasites with the haplotype CVMNK were detected in 19 individuals prior to treatment (20.43%; 95% C.I. 5.14-18.5%). In these patients, CQ treatment rapidly selected CVIET parasites, and this haplotype overwhelmingly dominated the parasite population in each individual by day 2 after treatment. CONCLUSIONS: Such rapid intra-host selection of particular genotypes after the introduction of drug will cause frequent misidentification of parasite genotypes present in the starting population. This will have a potentially serious confounding effect on clinical trials which employ PCR-corrected estimates of treatment failure, as resistant parasites below the detection threshold in the pre-treatment sample can be erroneously classified as "new" infections during follow-up, over-estimating drug efficacy.