Molecular analysis of mutations at the hypoxanthine-guanine phosphoribosyltransferase (HPRT) locus in peripheral blood T-lymphocytes can provide information on mechanisms of somatic in vivo mutation in populations exposed to exogenous carcinogens and in individuals with inherent susceptibility to cancer and other diseases. To study possible mutational changes associated with smoking as a risk factor for lung cancer, we analyzed HPRT mutations in T-cells of newly diagnosed, nonsmoking and smoking lung cancer patients before treatment. Reverse transcriptase polymerase chain reaction (RT-PCR) and DNA sequencing methods were used to identify 146 independent mutations, 73 each from 32 nonsmoking and 31 smoking cases. In 35 T-cell mutants, the HPRT cDNA showed loss of an entire exon, indicating a splicing mutation. Among the remaining 111 fully characterized mutations in the coding region, single base pair (bp) substitutions predominated with 79% (48/61) in nonsmokers and 90% (45/50) in smokers. Frameshift and small deletion (1-24 bp) mutations were found in 18 mutants. The distribution of base pair substitutions was nonrandom, with significant clustering at previously identified hotspot positions 143, 197 and 617 in the HPRT coding sequence (P< or =0.008). One additional hotspot, GC-->TA at position 606, was observed only in smokers (P=0.006). The frequency of GC>TA transversions was higher in smokers (13%) than in nonsmokers (6%). Conversely, smokers had a lower frequency of GC>AT transitions (24%) than nonsmokers (35%). This smoking-associated shift of the HPRT mutational spectrum, although not statistically significant, is consistent with the in vitro mutagenicity of benzo(a)pyrene (BaP), a prominent carcinogen of tobacco smoke, and with known differences in the TP53 mutational spectrum in lung tumors of smokers and nonsmokers. Among nonsmokers, the HPRT mutational spectra in healthy population controls and lung cancer patients were similar, but there was a marginally significant difference (P=0.07) in the distribution of base pair substitutions between smoking controls and patients. These results suggest that (i) general mechanisms of somatic mutagenesis in individuals with possible predisposition to cancer (e.g. nonsmoking lung cancer patients) are not different from those in normal healthy individuals, and (ii) the HPRT gene in T-cells is a useful reporter locus for smoking-associated somatic in vivo mutations occurring early in lung cancer development.