Significant variability in critical study parameters such as tumor incidences and survival, increasing tumor incidence and decreasing survival in common toxicity test models, and agent-induced changes in body weight (BW) and BW distribution all generate concern about the reproducibility, consistency, and equity of chronic toxicity tests used in regulation. These concerns have led to suggestions to control BW in chronic tests by the modulation of dietary intake without inducing malnutrition [dietary control (DC)] thereby minimizing tumor and survival variability both between and within studies. Evaluating the reports of the best controlled set of chronic experiments, the National Toxicology Program bioassay series, from studies initiated from 1981 to 1990, there is an increase in tumor incidence, especially liver tumors, with a consistent increase in BW. The studies are classified as to whether normal or aberrant BW growth curves occur. When the studies with normal growth curves are considered, the variance in the BW at 12 mo on test (BW12) can account for over 50% of the variance in liver tumor incidence. Additional stratification by study type, which alter tumor prevalences, as well as appreciation of housing effects [group housing decreases survival (in male mice) and induces tumors in males and females when compared to individual housing], further increase the strength of the correlations, accounting for up to 90% of the variance seen in tumor incidences. These updated analyses further support the hypothesis that it is the BW variation that is resulting in much of the variability seen in tumor incidences and refine the suggestions for the BW curves used as the desired targets for DC.