A review was presented of exposure related classification variables in epidemiological studies of occupational hazards, and suggested criteria for selection of exposure indicators to obtain maximal information were outlined. The basis of classification criteria included description of evidence, internal validity, external validity and comparability of results with other available evidence. Exposure classification required selection of a study population to achieve clear delineation between exposure and nonexposure, availability of good quality data and other reliable information, no unresolvable confounding factors and a large enough population for powerful analysis. Important to causal associations were reasonable time relationship of exposure to disease, dose response relationship and internal consistency. Study hypotheses required accurate formulation with reference to specific agents and their characteristics and possible modifiers or confounders. Often, the primary restricting factor for exposure assessment was availability and quality of data on work histories and exposure levels. Exposure stratification variables could include proxy of qualitative exposure; qualitative exposure to an agent; ordinal, semiquantitative or quantitative exposure; time relationship of exposure or its proxy; and adjusted and joint exposure indicators. Advantages and disadvantages were discussed along with possibilities for exposure misclassification. Model calculations were used to demonstrate mutual relationships of exposure indicators. Exposure misclassification or other mechanisms could cause serious biasing of risk and exposure response relationships when proxies of exposure or target dose were used. Good prediction of outcome was likely for indicators adjusted for main confounders and modifiers and those addressing exposure pattern or mode, when the effect of such factors was strong.