Urinary banzo[a]pyrene (BaP) metabolite levels were compared to human environmental exposure to BaP through inhalation and dietary ingestion to assess the predictive validity of the exposure biomarker. These measurements were made for 14 adult volunteers over 14 consecutive days, once during summer/fall, again during winter periods. Based on personal air monitoring, median potential inhalation doses of 11.0 and 2.3 ng/day were estimated for the winter and summer/fall studies, respectively. A median potential ingested dose of 176 ng/day, estimated from "duplicate plate" sampling, exceeded inhalation by 6- and 122-fold for the winter and summer/fall studies, respectively. "Total" urinary BaP metabolites were measured using a published "reverse metabolism" (BaP) method of analysis. Median rates of urinary BaP metabolite elimination for the winter and summer/fall studies were 121 and 129 ng/day, respectively. The changes in inhaled and ingested potential doses were regressed on the change in urinary metabolite elimination from week 1 to week 2 to test the predictive validity of the biomarker measurement. The regression was statistically significant (r = 0.620, p = 0.015, n = 25) when body weight was included and two extreme values were removed. Consistent with the exposure measurements showing diet as the dominant route of exposure, most of the variation in urinary metabolite elimination was explained by the ingested dose. It is concluded that the measurement of urinary BaP by "reverse metabolism" is qualitative and of marginal predictive validity as an exposure biomarker due to the method's low recoveries and the large unexplained variance.