The p53 gene is frequently mutated in lung tumors, and mutations may be caused by both polycyclic aromatic hydrocarbons (PAHs) and nitrosamines found in tobacco smoke. The two major forms of lung cancer, adenocarcinoma (AC) and squamous cell carcinoma (SCC), are known to differ in the proportion of tumors exhibiting p53 mutation, and may also differ in the mutational spectra produced. Previous studies comparing p53 mutational spectra between AC and SCC of the lung have been limited by small sample size. We examined p53 mutations in exons 5-8 in 202 cases of AC and 82 cases of SCC from smoking lung cancer patients in the Western Pennsylvania region. The percent of cases with p53 mutation was significantly lower in ACs (40/202, 20%) compared to SCCs (29/82, 35%, P=0.006). The proportion of mutations present that were G to T transversions was not significantly different between the two tumor types (52% of p53 mutations in AC compared to 32% in SCC). G to A transitions either did not differ in frequency in the two types of lung cancer (20% of mutations in AC and 24% of mutations in SCC). A distinct spectrum was observed, however, in the p53 mutation pattern in the two types of lung cancer. ACs showed a strong preference for a mutational hotspot at codons 248 and 249, while squamous cell tumors showed mutational events spread throughout exons 5-8, with a preference for codon 267. Mutations at codon 267 in SCC were all C to T transitions that occurred at CpG sites. Both tumor types demonstrated preferential mutation of the non-transcribed strand (100% of all G to T transversions and 55% of the G to A transitions). These results suggest that p53 mutations in both types of lung tumors may arise from adduction by both PAHs and nitrosamines. Mutations arising in ACs appear selectively in regions of p53 that produce more rigid proteins, suggesting a drastic change in p53 function is needed to result in ACs, while less constrained changes in p53 function can result in SCCs. Mutation in p53 was not found to be related to patient survival in this group of patients, while tumor size and degree of differentiation were poor survival predictors.