Environmental, demographic, and entomologic variables were analyzed by logistic multiple regression analysis for their association with the likelihood of being seropositive for Trypanosoma cruzi in three highly infested rural villages of northwest Argentina. The prevalence of seropositivity for T. cruzi, as determined by the composite results of three serologic tests, was 34% among 338 persons in 1992. The strongest positive predictors of the adjusted odds of being infected were the household number of dogs, the density of T. cruzi-infected Triatoma infestans in bedroom areas, and each person's age. Dwellers from houses with roofs made completely or partly with a grass called simbol, or which used insecticides rudimentarily and nonsystematically, had a significantly lower odds of being seropositive for T. cruzi than residents from other types of dwellings. The adjusted odds of infection also increased with the number of T. cruzi-infected dogs or cats and the presence of chickens in bedroom areas. No significant effects on the adjusted odds of infection of a community-wide deltamethrin spraying carried out in one of the villages seven years before were detected. Socioeconomic indicators, such as domiciliary area, and numbers of corrals and livestock, were inversely related to being infected. Our study identified several manageable variables suitable for control actions, most of them not examined before in univariate or multivariate analyses. Environmental management based on low-cost housing with appropriate local materials and removal of domestic animals from domiciliary areas have a crucial role to play in the control of Chagas' disease in rural areas.