Twelve carcinogenicity experiments in NMRI mice from two different breeders were started at three different locations between 1974 and 1979. Tumors occurring in untreated control mice were compiled on the basis of histopathological reports. Of the 629 male and 627 female mice at the beginning of the studies, 614 males and 591 females were evaluated. The remaining animals were unavailable due to autolysis. The histopathologic diagnoses were performed by a team of 5 pathologists. A total of 1,220 tumors was seen (585 in males and 635 in females), of which 527 were classified as malignant (218 in males and 309 in females). About 56% of all tumors were located in the lungs and lymphoreticular system (LRS). Average frequencies of primary tumors: lungs 28.7%, LRS 28.2%, ovaries 21.3%, uterus 4.6%, vascular tissue 4.5%, mammary glands 3.9%, liver 3.2%, Harderian glands 3.2%, testes 3.1%, adrenals 3.1%, stomach 2.9%, pituitary 2.2%, endocrine pancreas 1.2%, thyroid 1.0%, Cowper's glands 1.0% and various other organs having tumor frequencies of less than 1%. This wide spectrum of spontaneous tumors with comparatively low frequencies renders NMRI mice a suitable strain for carcinogenicity studies. Male animals exhibited strikingly more tumors of the lungs, liver, Harderian glands, adrenals, endocrine pancreas and intestine, while LRS and pituitary tumors were predominant in females. Some tumor frequencies seem to be influenced by different husbandry conditions and different stocks. However, even under nearly identical husbandry conditions, the general frequency of benign and/or malignant tumors and of particular tumors varied markedly from one study to the next. Differences in mortality among studies had no pronounced effect on the range of tumor frequencies. Due to this biological variability in animal material, an interpretation of tumor frequencies in carcinogenicity studies necessitates an in-depth knowledge of the variability of tumor data for this mouse strain under similar husbandry conditions.