Human respiratory health hazards for people working in livestock confinement buildings have been recognized since 1974. However, before comprehensive control programs can be implemented, more knowledge is needed of specific hazardous substances present in the air of these buildings, and at what concentrations they are harmful. Therefore, a medical epidemiological and exposure-response study was conducted on 207 swine producers using intensive housing systems (108 farms). Dose-response relationships between pulmonary function and exposures are reported here. Positive correlations were seen between change in pulmonary function over a work period and exposure to total dust, respirable dust, ammonia, respirable endotoxin, and the interactions of age-of-producer and dust exposure and years-of-working-in-the-facility and dust exposure. Relationships between baseline pulmonary function and exposures were not strong and therefore, not pursued in this study. The correlations between exposure and response were stronger after 6 years of exposure. Multiple regression models were used to identify total dust and ammonia as the two primary environmental predictors of pulmonary function decrements over a work period. The regression models were then used to determine exposure concentrations related to pulmonary function decrements suggestive of a health hazard. Total dust concentrations > or = 2.8 mg/m3 were predictive of a work period decrement of > or = 10% in FEV1. Ammonia concentrations of > or = 7.5 ppm were predictive of a > or = 3% work period decrement in FEV1. These predictive concentrations were similar to a previous dose-response study, which suggested 2.5 mg/m3 of total dust and 7 ppm of NH3 were associated with significant work period decrements. Therefore, dust > or = 2.8 mg/m3 and ammonia > or = 7.5 ppm should be considered reasonable evidence for guidelines regarding hazardous exposure concentrations in this work environment.