Studies using the N-nitrosomethylurea mammary tumor model indicate that the tumor-promoting effect of dietary fat is dependent on both qualitative and quantitative factors. Rats were fed diets containing either safflower, corn, or olive oil at either 23 (HF) or 5% (LF) in the diet (w/w). Coconut oil was fed to one group at 23% (w/w). It was found that animals fed HF diets rich in linoleic acid, such as safflower and corn oil, exhibited increased incidence and decreased latent period compared with either their LF counterparts or animals fed HF diets rich in oleic acid (olive oil) or medium-chain saturated fatty acids (coconut oil). Analysis of tumor lipid fatty acid content indicated that tumor-neutral lipids reflected the diet whereas tumor phospholipids did not. Moreover, tumor prostaglandins (PGE2) were high in the two high-incidence groups (HF safflower and corn) and low in the two low-incidence groups (HF olive and coconut). These results indicate that HF intake is a necessary, but not a sufficient, condition for mammary tumor promotion, and that the proportion of essential polyunsaturates vis a vis monounsaturates and saturates is a critical determinant of the fat effect. In addition, our studies suggest that alterations in the metabolism of linoleic acid to (prosta . . .) prostaglandins may underlie the fat effect. Dose-response studies in the same model, using four different levels of corn oil, suggest that instead of a linear relationship with respect to tumor incidence, there appears to be a threshold lying between 20 and 33% fat as calories, above which tumor promotion is manifested and below which it is not.