Acrylonitrile (ACN) has been tested for carcinogenicity by various routes in a number of rat strains. At relatively high levels of administration (e.g. 500 ppm in the drinking-water) there were statistically significant increases in microscopically detectable primary brain tumours, which were difficult to classify. In a further study of ACN-induced brain tumours, ACN was administered to groups of 50 male and 50 female F-344 rats from 6 wk of age at levels of 0, 100 and 500 ppm in the drinking-water. A fourth group of 300 rats (147 males, 153 females), was also given 500 ppm ACN. Neurological signs were observed in 0, 4, 16 and 29, respectively, of the rats in these four groups within 12-18 months. Among the treated animals, females died slightly earlier than males. Few controls of either sex had died by month 18, but, apart from those killed for tumour donation, a high proportion of the rats in the 500-ppm groups had died by that time. Of the 49 brain tumours found in rats exposed to 500 ppm ACN, 11 were only detectable microscopically, 28 were 1-5 mm in diameter and 10 were greater than 5 mm. Despite this variation in size, all the tumours were similar in cellular and architectural features. They were densely cellular, with occasional areas of focal necrosis, and were infiltrative at the margins. They were negative for glial fibrillary acidic protein (GFAP). Ultrastructurally, the tumour cells showed intermingling cytoplasmic processes but no glial filaments and no neurosecretory granules or specialized cell contacts. Samples of tumour tissue were successfully grown in culture, but transplantation of samples from these cultures (observed for up to 12 wk) was unsuccessful. However, a direct intracerebral transplantation from a large tumour was successful.