Air monitoring and biological monitoring were discussed as surrogates of target dose and compared in terms of exposure fluctuations over time and in terms of pharmacokinetic factors as potential modifiers. Target dose on biological receptors was rarely accessible for measurement, so biological exposure indicators were used to estimate this. Such indicators gave more accurate measures of absorbed dose and accounted for differences in fate of absorbed chemicals in the body. For most chemicals of interest, only retrospective air monitoring data were available. These monitoring methods were compared for perchloroethylene (127184) exposures in dry cleaning and industrial mercury (7439976) exposures. Statistically significant relationships were obtained, but wide scatter was noted. This may have resulted from long term effects seen in biological measures but not in single day air measurements, other routes of absorption or interindividual differences in distribution and metabolism. Grouping into exposure categories produced a high rate of misclassification. Air sampling could detect long term but not short term fluctuations reliably, while biological half lives influenced measurement of fluctuations by biological monitoring. Half life damped exposure variability for indicators with long half lives but was subject to individual variability. Statistically, air monitoring was generally better for indicators with half lives below 10 hours and biological monitoring was generally better for those with longer half lives. Modifiers of such measures included skin absorption, physical workload with respect to pulmonary ventilation, aerosol solubility and particle size and physiological fate of chemicals. A predictive model for skin absorption compared to respiratory exposure was presented. The authors suggest that when air monitoring is considered the appropriate method, biological monitoring could be used on a smaller scale to estimate magnitudes of potential biases.