The prevalence of eating disorders and obesity in western societies is epidemic and increasing in severity. Preclinical research has focused on the development of animal models that can mimic the maladaptive patterns of food intake observed in certain forms of eating disorders and obesity. This study was aimed at characterizing a recently established model of palatable diet alternation in female rats. For this purpose, females rats were fed either continuously with a regular chow diet (Chow/Chow) or intermittently with a regular chow diet for 2 days and a palatable, high-sucrose diet for 1 day (Chow/Palatable). Following diet cycling, rats were administered rimonabant (0, 0.3, 1, 3 mg/kg intraperitoneally) during access to either palatable diet or chow diet and were assessed for food intake and body weight. Finally, rats were pretreated with rimonabant (0, 3 mg/kg, intraperitoneally) and tested in the elevated plus maze during withdrawal from the palatable diet. Female rats with alternating access to palatable food cycled their intake, overeating during access to the palatable diet and undereating upon returning to the regular chow diet. Rimonabant treatment resulted in increased chow hypophagia and anxiety-like behavior in Chow/Palatable rats. No effect of drug treatment was observed on the compulsive eating of palatable food in the diet-cycled rats. The results of this study suggest that withdrawal from alternating access to the palatable diet makes individuals vulnerable to the anxiogenic effects of rimonabant and provides etiological factors potentially responsible for the emergence of severe psychiatric side-effects following rimonabant treatment in obese patients.