OBJECTIVE: To present a sensitivity analysis of the most widely used means of estimating lifetime occupational exposure proportion (LOEP) and their respective impacts on LOEP and population-attributable fraction (PAF) estimates. METHODS: A French population-based sample with full job history (N = 10,010) was linked with four Matgéné job-exposure matrices: flour, cement, silica and benzene. LOEP and the 95% confidence interval were estimated using four methods: the maximum exposure probability during the career (Proba_max), two methods subdividing careers into job-periods (job-period_M1, job-period_M2) and one into job-years (job-year). To quantify differences between methods, percentages of variation were calculated for proportion values and PAF, and compared with published results for France using cross-sectional proportion multiplied by a factor. RESULTS: For each agent, LOEP estimated from the maximum probability during the career (Proba_max) was consistently lower than proportion taking account of job-periods or job-years. LOEP on Proba_max for flour, cement, silica and benzene were, respectively, 4.4% 95% CI (4.0-4.7), 4.3% (3.9-4.6), 6.1% (5.7-6.5) and 3.9% (3.6-4.2). Percentage of variation ranged from 0 to 55.8% according to the agent. The number of cancer cases varied by a twofold factor for exposure to silica and lung cancer and by a fourfold factor for exposure to benzene and acute myeloid lymphoma. CONCLUSIONS: The present study provides a description of several LOEP estimation methods based on exposure assessment over the entire career and describes their impact on PAF. For health monitoring purposes, we recommend to report a range of LOEP with low and high estimates obtained using job-periods (job-period_M1 and job-period_M2).< Réduire