This study was designed to identify and describe smoking-related differences in dietary and nutritional factors that are potential independent predictors of cancer risk or effect modifiers or confounders of tobacco-cancer relationships. Data were obtained from a large hospital-based case-control study which was designed to estimate the cancer risk from various tobacco products and consisted of 465 male and 300 female incident lung cancer cases and 870 male and 556 female hospitalized patient controls matched on sex and age (+/- 5 years). Nutritional data were analysed as log-transformed frequencies of thirty food items, nine factor scores generated to describe overall patterns of dietary intake, and estimated daily nutrient scores for fat, vitamin A, fibre, and cholesterol. In general, the dietary habits of ex-smokers more closely resembled those of never-smokers than those of current smokers. We found that after controlling for case-control status, education, alcohol consumption, and age, there were many more significant differences in nutritional exposures by smoking status than could be explained merely by chance. For both sexes we observed significantly increased consumption of fruits and higher vitamin A and fibre scores in non-smokers compared to current smokers (for any smoking vs non-smoking comparison the P-value was always less than 0.002, 0.01, and 0.007, respectively). A similar but weaker relationship was observed for high-fat, sweet foods such as ice cream. An inverse association, also of smaller magnitude, was found for other high-fat foods items. Implications for further study and strengths and weaknesses of the current study are discussed.