OBJECTIVES: To explore the effects of various factors related to the industry, the contaminant, and the period and type of sampling on long term declining trends in occupational exposure. METHODS: Linear regression analyses were used to assess the relation between reductions in exposure and geographical location, industrial sector, type of contaminant, type of monitoring, carcinogenic classification, calendar period, duration of sampling, and number of reductions in the threshold limit value during the sampling period. Both univariable and multivariable models were applied. RESULTS: Based on univariable analyses, the findings suggest that exposures declined more rapidly in manufacturing than in mining, more rapidly for aerosol contaminants than for vapours, and more rapidly when biological, rather than airborne, monitoring was conducted. Exposures collected more recently (first year of sampling in 1972 or later) fell more rapidly than exposures first evaluated during earlier periods. Irrespective of when the data were collected, the results also suggest that the longer the duration of sampling the slower the rate of decline. Taken together, we found that characteristics related to the contaminant, the industry, the sampling period, and the type of sampling explained a substantial proportion of the variability for exposures evaluated before 1972 (R2 = 0.78) and for sites evaluated both before and after 1972 (R2 = 0.91), but explained essentially no variation for data gathered exclusively after 1972 (R2 = 0.04). CONCLUSIONS: By identifying factors that have affected the rates of reduction in a consistent fashion, the results should guide investigators in estimating historical levels when studies assessing exposure-response relations are carried out.