This publication is the fourth in a series of safety evaluations performed by the Expert Panel of the Flavour and Extract Manufacturers' Association (FEMA). In 1993, the Panel initiated a comprehensive program to re-evaluate the safety of more than 1700 GRAS flavouring substances under conditions of intended use. In this review, scientific data relevant to the safety evaluation of trans-anethole (i.e. 4-methoxypropenylbenzene) as a flavouring substance is critically evaluated by the FEMA Expert Panel. The evaluation uses a mechanism-based approach in which production of the hepatotoxic metabolite anethole epoxide (AE) is used to interpret the pathological changes observed in different species and sexes of laboratory rodents in chronic and subchronic dietary studies. Female Sprague-Dawley rats metabolize more trans-anethole to AE than mice or humans and, therefore, are the most conservative model for evaluating the potential for AE-induced hepatotoxicity in humans exposed to trans-anethole from use as a flavouring substance. At low levels of exposure, trans-anethole is efficiently detoxicated in rodents and humans primarily by O-demethylation and ω-oxidation, respectively, while epoxidation is only a minor pathway. At high dose levels in rats, particularly females, a metabolic shift occurs resulting in increased epoxidation and formation of AE. Lower activity of the 'fast' acting detoxication enzyme epoxide hydrolase in the female is associated with more pronounced hepatotoxicity compared to that in the male. The continuous intake of high dose levels of trans-anethole (i.e. cumulative exposure) has been shown in dietary studies to induce a continuum of cytotoxicity, cell necrosis and cell proliferation. In chronic dietary studies in rats, hepatotoxicity was observed when the estimated daily hepatic production of AE exceeded 30mg AE/kg body weight. In female rats, chronic hepatotoxicity and a low incidence of liver tumours were reported at a dietary intake of 550mg trans-anethole/kg body weight/day. Under these conditions, daily hepatic production of AE exceeded 120mg/kg body weight. Additionally, neither trans-anethole nor AE show any evidence of genotoxicity. Therefore, the weight of evidence supports the conclusion that hepatocarcinogenic effects in the female rat occur via a non-genotoxic mechanism and are secondary to hepatotoxicity caused by continuous exposure to high hepatocellular concentrations of AE. trans-Anethole was reaffirmed as GRAS (GRASr) based on (1) its low level of flavour intake (54μg/kg body weight/day); (2) its metabolic detoxication pathway in humans at levels of exposure from use as a flavouring substance; (3) the lack of mutagenic or genotoxic potential; (4) the NOAEL of 120mg trans-anethole/kg body weight/day in the female rat reported in a 2+-year study which produces a level of AE (i.e. 22mg AE/kg body weight/day) at least 10,000 times the level (0.002mg AE/kg body weight/day) produced from the intake of trans-anethole from use as a flavouring substance; and (5) the conclusion that a slight increase in the incidence of hepatocellular tumours in the high dose group (550mg trans-anethole/kg body weight/day) of female rats was the only significant neoplastic finding in a 2+-year dietary study. This finding is concluded to be secondary to hepatotoxicity induced by high hepatocellular concentrations of AE generated under conditions of the study. Because trans-anethole undergoes efficient metabolic detoxication in humans at low levels of exposure, the neoplastic effects in rats associated with dose-dependent hepatotoxicity are not indicative of any significant risk to human health from the use of trans-anethole as a flavouring substance. Copyright (C) 1999 Elsevier Science Ltd.