Retrospective assessment of asbestos exposure--I. Case-control analysis in a study of lung cancer: efficiency of job-specific questionnaires and job exposure matrices. International Journal of Epidemiology 1993; 22 (Suppl. 2): S83-S95. In a lung cancer study in Northern Germany the asbestos exposure assessment obtained from detailed supplementary questionnaires (SQ) was compared to the assessment obtained by the application of two job exposure matrices (JEM) to the job history. The study includes 391 incident male cases and the same number of controls from the general population, matched by sex, age, and region. Almost 16% of the subjects are considered as never having been exposed to asbestos and 24% are classified as certainly exposed according to both of the JEM, the corresponding percentages of the SQ-method being 68% and 32%. Sixty percent are considered by the JEM as having an intermediate probability of exposure. In general, the agreement between the different methods is better when the exposure definition is restricted to definite exposures, the corresponding Kappa statistic being 0.67 for the comparison between both JEM and 0.44/0.39 between the JEM and the SQ. The positive agreement between SQ and JEM (reference) increases from 12% and 17% for subjects with low probability of exposure to 61% and 69% for those with definite exposures. A ranking according to cumulative exposure shows some dose-response relationship for each of the three methods with a smoking-adjusted OR in the highest category of 1.85 (95% CI: 1.01-3.38) for the SQ method and 2.47 (95% CI: 1.40-4.37) and 2.94 (95% CI: 1.73-4.99) for the two JEM. The results are discussed with reference to the conceptual differences between the methods and their potential scope in future studies.