Eleven two-year toxicity/carcinogenicity experiments in control Wistar TNO/W.70 rats were started between 1973 and 1976. Tumors occurring were compiled on the basis of histopathological reports. Of the 1,000 male and 1,000 female rats at the beginning of the studies, 962 males and 968 females were evaluated. The remaining animals were unavailable due to autolysis or early death (before day 400). The histopathologic diagnosis for the eleven reports were performed by five different pathologists or groups of pathologists. A total of 827 tumors (375 in males and 452 in females) was seen in 686 rats (303 males, 383 females). 183 tumors (67 in males, 116 in females) were classified as malignant. 719 (86.9%) of all tumors and 120 (65.6%) of all malignant tumors were located in endocrine organs (pituitary, thyroid, adrenal, pancreatic islets) or in genital organs (testes, epididymides, seminal vesicles, ovaries, uterus, mammary gland). Average incidences of primary tumors found in the following organs were: pituitary 20.0%, thyroid 9.6%, uterus 9.2%, adrenals 4.1%, mamma 3.2%, testes 2.9%, ovaries 1.8%, skin 1.4% and 17 other organs each with a tumor frequency below 1%. This wide spectrum of spontaneous tumors with a mostly low incidence makes the Wistar rat a preferred strain for carcinogenicity studies. Despite nearly identical husbandry conditions, a marked variability was observed among experiments regarding the general incidence of benign and/or malignant tumors, particular tumors, and number of tumor-bearing animals. Differences in mortality among experiments had no pronounced effect on the variability of tumor incidences. Different evaluation criteria used by different pathologists did not appear to be a major cause of the observed variability with the probable exception of lesions in endocrine organs. The major cause would seem to be the biological variability of the animal material. The observed marked variability of tumor incidences between experiments would indicate that a considerable natural variability may also occur between groups of one carcinogenicity study. Therefore, information on historical control tumor data from the same rat strain kept under similar conditions is needed to interpret the incidences of tumors in carcinogenicity experiments.