Estimating the carcinogenic potential of exposure to diesel-engine exhaust particulates (DEPs) is problematic. In rats, high concentrations of DEPs (> 1,000 microg/m(3)) inhaled over a lifetime result in excess lung tumors. However, data for rats exposed to DEP at concentrations not associated with lung overload are consistent with no tumorigenic effect. Individual rat studies have only a limited number of exposure groups; therefore, we combined the tumor data from eight chronic inhalation studies in a meta-analysis. Statistical analysis identified a threshold of response between 200 and 600 microg/m(3) average continuous lifetime exposure, consistent with biological-effect thresholds reported by other investigators. Our exposure-response analysis of all rats with < 600 microg/m(3) average continuous lifetime exposure found no tumorigenic effect of DEP in these rats. When we evaluated all rat studies, accounted for a threshold and for inhomogeneity between experiments, and expressed the results in terms of human unit risk (UR), we found a negative maximum-likelihood human UR of -32 (times) 10(-6) per microgram per cubic meter (microg/m(3)), but this was not statistically significantly different from zero. Extrapolating the rat upper 95th percentile confidence limit to humans gave an upper-bound human UR of 9.3 (times) 10(-6) per microg/m(3)]. This upper-bound human UR, derived from all of the data points (including 1,087 animals below the estimated threshold and 1,433 in the control groups), falls entirely below the range of estimates derived from lung-overloaded rats or from epidemiology of railroad workers. Our meta-analysis of the low-exposure data in rats does not support a lung cancer risk for DEP exposure at nonoverload conditions. Average ambient concentrations of DEP (0-3 microg/m(3)) are < 1% of the concentration associated here with a threshold of tumor response in the rat bioassay.