The paper addresses the problem of relating measurements of polycyclic organic matter (POM) in source emissions to the potential hazard associated with the total POM in the sample measured. Although uncertainties exist in these quantitative measurements of POM, a more serious problem concerns the relation of hazard to a given mass and composition of POM. As used here, hazard is defined as the quantity that may be used to indicate the relative environmental problems of various source emissions. For each individual type of POM source, both benzo(a)-pyrene--B(a)P--and total POM provide excellent indicators of hazard, evaluated by a weighted sum of concentrations. The method of using a linear combination of masses of the individual POM compounds to model hazard due to POM is shown to be appropriate under certain conditions, and to follow directly from a simple probabilistic model of the physical and chemical events leading up to a response event. The results suggest that, in most cases, total POM measurements provide somewhat more consistent hazard prediction than those of B(a)P. In view of this, the total POM indicator should be more useful in predicting hazard for sources for which the compound-specific composition of POM emitted is not well known.