Two structural descriptors were chosen as possible model descriptors for predicting subcooled-liquid vapor pressures of polychlorinated biphenyls (PCBs), chlorophenols (CPs), chlorobenzenes (CBs) and alkylphenols and predictive power of the two descriptors are discussed. The Characteristic Root Index (CRT) and third order cluster-type molecular connectivity index ((3) chi(c)) were calculated exclusively on the basis of information readily available for all chemicals. To apply the regression model to a wide range of chemicals multiple linear regression has been tried, however, collinearity was observed between the two structural descriptors. Subcooled-liqiud vapor pressures obtained by using the best correlative equation with the CRI for 54 PCB congeners, all chlorophenols and chlorobenzenes and several alkylphenols have been reported. (C) 1997 Elsevier Science Ltd.