BACKGROUND. Antimalarial use is a key factor driving drug resistance and reduced treatment effectiveness in Plasmodium falciparum malaria, but there are few formal, quantitative analyses of this process. METHODS. We analyzed drug usage, drug failure rates, and the frequencies of mutations and haplotypes known to be associated with drug resistance over a 12-year period (1991-2002) in a site in Papua New Guinea. This period included 2 successive treatment policies: amodiaquine (AQ) or chloroquine (CQ) from 1991 through 2000 and their subsequent replacement by sulfadoxine-pyrimethamine (SP) plus AQ or SP plus CQ. RESULTS. Drug use approximated 1 treatment per person-year and was associated with increasing frequencies of pfcrt and pfmdr1 mutations and of treatment failure. The frequency of pfdhfr mutations also increased, especially after the change in treatment policy. Treatment failure rates multiplied by 3.5 between 1996 and 2000 but then decreased dramatically after treatment policy change. CONCLUSIONS. With high levels of resistance to CQ, AQ, and SP, the deployment of the combination of both drugs appears to increase clinical effectiveness but does not decelerate growth of resistance. Our estimates of mutation and haplotype frequencies provide estimates of selection coefficients acting in this environment, which are key parameters for understanding the dynamics of resistance.