The antioxidant food preservative butylated hydroxyanisole (BHA) was tested in an initiation-promotion protocol in which male F344 rats (6 wk old), 27 per group, were gavaged with a single dose of 200 mg N-methyl-N'-nitro-N-nitrosoguanidine (MNNG)/kg. After 3 wk on control diet, test diets containing 0, 60, 300, 1000, 3000, 6000 or 12,000 ppm BHA were fed until termination of the experiment at approximately 110 wk, at which time most animals had died with stomach tumours. MNNG caused a high incidence of tumours in the glandular stomach and forestomach of all groups. Administration of 12,000 and 6000 ppm BHA, but not 3000 ppm or lower doses, caused statistically significant increases in the time-related incidence of MNNG-induced forestomach tumours as analyzed by life table analysis. BHA had no effect on the incidence of tumours in the glandular stomach or oesophagus. Tumour incidences in other organs were not related to BHA dose. No increase in hyperplasia in the oesophagus was evident in the high-dose BHA-treated animals compared with the MNNG-only group. This study provides corroboration that BHA affects only forestomach tumorigenesis and that the dose for enhancement of tumorigenesis is at least 1500-fold greater than human exposure.