In occupational mortality studies, the information on the population at risk is frequently not available. In these circumstances, the rate ratio of interest can be assessed through the mortality odds ratio (MOR), given appropriate selections of reference occupation(s) and reference disease(s). The reference occupation should be chosen with a view to three aspects of validity: (1) comparability of effects, i.e., identity of mortality effects between the index and the reference occupation apart from the effect of the exposure under study; (2) comparability of contrasted populations, i.e., same job entry and exit factors and similar health promotion programmes when related to the mortality under study; and (3) comparability of mortality information, i.e., same diagnostic and certification practices for the cause of death of interest between the contrasted populations. The reference causes of death must also satisfy these requirements with the additions that the exposure under study have no effect on the risk of reference causes of death, and that the accuracy of information about the compared occupations be the same as in the context of the index cause of death. Throughout the discussion, an example of exploring cancer risks among lens manufacturing workers is used.