The therapeutic efficacy of chloroquine (CQ) and sulfadoxine-pyrimethamine (SP) was determined over a 4 year period (1998-2001) in Bobo Dioulasso, Burkina Faso, with an analysis of the risk factors associated to treatment failures to the 2 drugs. In total, 2008 children (6 months-15 years old) attending in 4 health centres (1 urban and 3 rural) were included in the study. Children were alternatively allocated to either CQ or SP The WHO 14-days in vivo field test was carried out. PCV was measured at day 0 and 14. CQ treatment failure was 24.4% (229/940), most of them being late failures. Between 1998 and 2001 a significant increase in CQ treatment failure (p < 0.001) was observed. SP showed a good efficacy with a total treatment failure of 4.4% (33/749). However; a significant increase of resistance to this drug (p=0.001) was also observed between 1998 and 2001. Among children with anaemia at day 0.85% (23/27) were no more anaemic by day 14 in the SP group, while in the CQ group the proportion was lower; 69% (27/39). However the difference between the two drugs was not significant (p > 0.1). Univariate analysis showed that the site, the age of children, the time of recruitment and the parasitaemia were significantly associated with CQ treatment failure. In the multivariate analysis these 4 variables remain significantly and independently associated with the risk of CQ treatment failure. After adjusting for the effect of the 3 other factors, the risk of treatment failure was reduced by half in rural area compared to urban area as well as in children of 5-15 years of age compared to those under 5. The risk of treatment failure was significantly increased in 2000-2001 (OR = 1.66, p < 0.05) as compared to the 2 previous years (1998-1999). It was also twice higher in children with parasitaemia > or = 16,000/microl than in those having a lower parasitaemia. For SP we have not observed such connexions with the univariate and multivariate analysis.