International health authorities have graded diesel motor emissions (DME) as probably cancerogenic in human beings. There are gaps in epidemiological evidence regarding exact exposure quantification, confounder control and the investigation of highly exposed populations. We investigated the association of DME and lung cancer mortality in a historical cohort study of 5,862 German potash miners who were followed from 1970 to 2001. Cumulative exposure (CE) was measured by representative concentrations of total carbon multiplied with exposure years from the mines' medical records. Exposure and smoking behavior were validated by interviews of 3,087 participants. We computed standardized mortality ratios (SMR, external comparison) and performed Cox regression (internal comparison). The relative risk estimates (RR) with 95%-confidence intervals were adjusted for age and smoking. Vital status and causes of death were confirmed for 98.1% of participants. Sixty-one lung cancer deaths occurred. SMR-analysis showed lower than expected lung cancer mortality (healthy-worker-effect). Internal comparisons revealed risk elevations from moderate to risk doubling depending on the exposure categories used (dichotomized: up to RR 1.43[0.67-3.03] for a CE of 4.90[mg/m(3)]*years as compared with less exposure; quintiles: RR 1.13[0.46-2.75], 2.47[1.02-6.02], 1.50[0.56-4.04] and 2.28[0.87-5.97] for a CE up to 2.04, 2.73, 3.90 and >3.90, respectively, as compared with the reference of <1.29[mg/m(3)]*years). Additional adjustment of length of follow-up leads to further RR increases and indicates healthy-worker-survivor-phenomena. The analyses of a sub-cohort (n = 3,335) with particularly accurate exposure measurement revealed a nonsignificant dose-response-relationship. Our results support an association of DME and lung cancer mortality.