In Papua New Guinea a 15-year mortality follow-up was carried out among 2026 highlanders and 1734 coastal dwellers on whom data on respiratory symptoms and signs, and lung function had been obtained by prevalence surveys in 1970/71. Survival status was established in 99%, of whom 9.8% had died, the rate being higher in the highland (12.2%) than in the coastal population (7.0%). All abnormalities indicative of chronic lung disease were associated with increased mortality though this was statistically significant only for chronic shortness of breath (SMR 195) and wheezing (SMR 183) in coastal females. In those aged 25 years or more there was a strong association between reduced lung function (FEV1, FVC, FEV1/FVC%) and mortality, after controlling for age and height. The associations observed were as strong as those reported from prospective studies in developed countries. The most striking difference was that in Papua New Guinea mortality showed no association whatever with tobacco smoking; this may be because the traditional type of tobacco and method of smoking it resembles pipe/cigar smoking in developed countries.