We report the results of an investigation conducted in 992 healthy control subjects (854 adults and 138 adolescents) and in 116 subjects with lung cancer (LC) for the purpose of detecting those individuals with a possible genetic predisposition to lung cancer. The test consists of the oral administration of 64 mumol of dextromethorphan (DMP) with collection of urine samples over the following 8-h period and urine assay of the drug (DMP) and its main metabolite, dextrorphan (DOP). The ratio of the urinary concentrations of DMP to those of DOP is called the metabolic ratio (DMP/DOP) and is inversely proportional to the DMP demethylation rate. The pattern of the metabolic ratio (Log10 DMP/DOP) allowed, using a maximum likelihood approach, the identification of three subpopulations in the 854 control subjects (adults): (1) probable homozygous extensive metabolizers with Log10 DMP/DOP < -1.74 (73.1 percent); (2) probable heterozygous intermediate metabolizers with Log10 DMP/DOP in the -1.74 to -0.40 range (22.3 percent); and (3) probable homozygous poor metabolizers with Log10 DMP/DOP > -0.4 (4.6 percent). Most of the patients with LC (89 percent) were probable homozygous extensive metabolizers. As the latter have a cancer risk that is 2.54-fold greater than that of intermediate metabolizers (95 percent confidence interval [CI]: 1.37 to 4.73) and 7.43-fold greater than that of poor metabolizers (95 percent CI: 1.01 to 54.5), their identification by means of the DMP test may be particularly useful for subjects exposed to environmental and occupational carcinogens. The phenotype test used is similar to that of the debrisoquin test, but presents the advantage that DMP is a widely used, harmless drug with a faster and simpler urinary assay procedure.