The goal of this study was to assess the impact of improved stoves, house ventilation, and child location on levels of indoor air pollution and child exposure in a rural Guatemalan population reliant on wood fuel. The study was a random sample of 204 households with children less than 18 months in a rural village in the western highlands of Guatemala. Socio-economic and household information was obtained by interview and observation. Twenty-four hour carbon monoxide (CO) was used as the primary measure of kitchen pollution and child exposure in all homes, using Gastec diffusion tubes. Twenty-four hour kitchen PM(3.5) was measured in a random sub-sample (n=29) of kitchens with co-located CO tubes. Almost 50% of the homes still used open fires, around 30% used chimney stoves (planchas) mostly from a large donor-funded programme, and the remainder of homes used various combinations including bottled gas and open fires. The 24-h kitchen CO was lowest for homes with self-purchased planchas: mean (95% CI) CO of 3.09 ppm (1.87-4.30) vs. 12.4 ppm (10.2-14.5) for open fires. The same ranking was found for child CO exposure, but with proportionately smaller differentials (P<0.0001). The 24-h kitchen PM(3.5) in the sub-sample showed similar differences (n=24, P<0.05). The predicted child PM for all 203 children (based on a regression model from the sub-sample) was 375 microg/m(3) (270-480) for self-purchased planchas and 536 microg/m(3) (488-584) for open fires. Multivariate analysis showed that stove/fuel type was the most important determinant of kitchen CO, with some effect of kitchen volume and eaves. Stove/fuel type was also the key determinant of child CO, with some effect of child position during cooking. The improved stoves in this community have been effective in reducing indoor air pollution and child exposure, although both measures were still high by international standards. Large donor-funded stove programmes need to aim for wider acceptance and uptake by the local families. Better stove maintenance is also required.