Since 2002, it is known that the probable human carcinogen acrylamide is present in commonly consumed carbohydrate-rich foods, such as French fries and potato chips. In this review, the authors discuss the body of evidence on acrylamide carcinogenicity from both epidemiological and rodent studies, including variability, strengths and weaknesses, how both types of evidence relate, and possible reasons for discrepancies. In both rats and humans, increased incidences of various cancer types were observed. In rats, increased incidences of mammary gland, thyroid tumors and scrotal mesothelioma were observed in both studies that were performed. In humans, increased risks of ovarian and endometrial cancers, renal cell cancer, estrogen (and progesterone) receptor-positive breast cancer, and oral cavity cancer (the latter in non-smoking women) were observed. Some cancer types were found in both rats and humans, e.g., endometrial cancer (observed in one of the two rat studies), but there are also some inconsistencies. Interestingly, in humans, some indications for inverse associations were observed for lung and bladder cancers in women, and prostate and oro- and hypopharynx cancers in men. These latter observations indicate that genotoxicity may not be the only mechanism by which acrylamide causes cancer. The estimated risks based on the epidemiological studies for the sites for which a positive association was observed were considerably higher than those based on extrapolations from the rat studies. The observed pattern of increased risks in the rat and epidemiological studies and the decreased risks in the epidemiological studies suggests that acrylamide might influence hormonal systems, for which rodents may not be good models.