Increased lung and nasal cancer risks have been reported in several cohorts of nickel refinery workers, but in more than 90% of the nickel-exposed workers that have been studied there is little, if any evidence of excess risk. This investigation utilizes human exposure measurements, animal data from cancer bioassays of three nickel compounds, and a mechanistic theory of nickel carcinogenesis to reconcile the disparities in lung cancer risk among nickel-exposed workers. Animal data and mechanistic theory suggest that the apparent absence of risk in workers with low nickel exposures is due to threshold-like responses in lung tumor incidence (oxidic nickel), tumor promotion (soluble nickel), and genetic damage (sulfidic nickel). When animal-based lung cancer dose-response functions for these compounds are extrapolated to humans, taking into account interspecies differences in deposition and clearance, differences in particle size distributions, and human work activity patterns, the predicted risks at occupational exposures are remarkably similar to those observed in nickel-exposed workers. This provides support for using the animal-based dose-response functions to estimate occupational exposure limits, which are found to be comparable to those in current use.