The toxic effects from exposure to halogenated hydrocarbons (HAs), which are produced in large amounts and used in a variety of applications, are well-known. Previously, QSARs for the toxicity of a series of HAs in vitro have been studied extensively. In this work, using a composite toxicity metric calculated from a set of five in vitro hepatotoxicity endpoints determined for 20 HAs, we find that QSARs derived using quantum descriptors calculated from the neutral HA species are statistically similar to QSARs calculated from HA metabolites. In most cases, QSARs derived using descriptors calculated from both neutral HAs and metabolites are statistically superior to those derived using either neutral-HA descriptors or metabolite descriptors. However, to properly utilize metabolite descriptors, multiple QSARs, each of which utilizes a set of HAs that form unique metabolites, must be derived and toxicity values calculated therefrom must be averaged. These average toxicity values agree better with experiment than those calculated from the neutral-HA QSARs.