We investigated the relationship between respiratory symptoms reported at one time and incidence of lung cancer the subsequent 30 years in an urban Norwegian population. A cohort of 19,998 persons, aged 15-70 years living in Oslo, was randomly selected for a respiratory survey in 1972. The response-rate was 89% and 17,670 respondents were followed up. The relationship between respiratory symptoms and lung cancer incidence was investigated separately for each symptom group, symptom score and sex, with adjustment for age, smoking habits and occupational exposure. Lung cancer developed in 352 persons (228 men and 124 women) during follow up. We found a significant positive association between the incidence of lung cancer and cough symptoms in both sexes, asthma-like symptoms among women and dyspnoea when walking uphill among men. The relative risk for lung cancer increased with the number of symptoms reported at baseline and was strongest the first decade and decreased with duration of follow up. This association was more pronounced for non-small cell lung cancer than for small cell lung cancer.