Chlorine gas, spreading from a train derailment, caused the deaths of 8 persons and the hospitalization of 23 with sublethal respiratory injuries. Twenty-five others had at least one sign of lower respiratory abnormality but were not hospitalized. One hundred thirteen who were examined for gas effects in the forty-eight hours after exposure, including 20 of 23 of those hospitalized and 21 of 25 of those not hospitalized but with respiratory abnormality, participated in follow-up studies. Probability of admission to hospital was related to distance from the spill, but by 3 wk after exposure there was no detectable difference in lung function relating to distance or apparent severity of injury. In 60 adults tested multiple times over the following 6 yr, longitudinal change in lung function showed expected differences related to smoking but none related to distance or severity of injury. The average annual change in FEV was -34 ml/yr in current smokers and -18 ml/yr in ex and never-smokers. The lack of a discernible chlorine effect in this cohort accords with the findings in most previous studies. Without pre-exposure measurements, a single, lasting reduction in lung function cannot be excluded, but there is no evidence for a persisting abnormal rate of decline.