A hypothesis-generating analysis of the role of diet on survival was conducted among a sample of 463 men and 212 women with histologically-confirmed lung cancer. Interview information was obtained from two population-based case-control studies of lung cancer conducted on the Island of Oahu, Hawaii, between 1979 and 1985. The interview consisted of a quantitative dietary history to assess the usual intake of foods 1 year prior to diagnosis, a complete tobacco history, and other demographic and lifestyle information. Records from the Hawaii Tumor Registry were reviewed for data on stage, histology, and follow-up status of these patients. A food group analysis showed a significant reduction in the risk of death with increasing consumption of all vegetables combined among women (P for trend = 0.03), but not among men. The covariate-adjusted median survival times for women from the highest to the lowest quartiles of vegetable intake were 33, 21, 15, and 18 months, respectively. The results also suggested an association of fruit intake and survival among women (P for trend = 0.02), although a similar effect was not found among men. Increased consumption of certain foods, such as tomatoes and oranges among men, and broccoli and, perhaps, tomatoes among women, appeared to improve survival. This exploratory analysis provides mixed indications that certain components of vegetables and fruits may prolong survival in lung cancer patients.