AbstractLubricant base oils are petroleum products that are predominantly derived from the vacuum distillation of crude oil. Various types of refinement can be employed during the manufacturing process, and evidence suggests that certain of the associated process streams produce skin cancer. Polycyclic aromatic compounds (PACs), some of which are considered as the causative agents, are removed, concentrated or chemically converted during the refinement process. In order to understand the effects of various types of refinement processes on carcinogenic potential, 94 oils were evaluated in the mouse epidermal cancer bioassay. This Exxon database is unique, because of the wide range of crude oils and processing histories represented. Seven processing history classifications are described, and conclusions concerning the impacts of each refinement process on dermal carcinogenicity are discussed.This research also included an evaluation of selected biological and chemical test methods for predicting carcinogenic potential. These included a modified version of the Ames test for mutagenicity, as well as analytical characterizations of the polycyclic aromatic structures in the oils. For classification purposes, a sample was considered to be carcinogenic if it resulted in the production of two or more tumor‐bearing animals (in test groups of either 40 or 50 animals). The modified Ames test was considered to be positive if the mutagenicity index was ≥ 2.0, and PAC analyses were similarly designated as positive or negative according to proposed guidelines. All of the alternative test methods showed similar agreement with dermal carcinogenicity bioassay data; concordance values were ≥ 80%. However, each test was incorrect in ca. 10%–20% of the cases evaluated. Thus, while these methods may have great value as screening tests, the results should not be used in preference to dermal carcinogenicity data for hazard identification or labeling p