For the assessment of the safety of food additives, data from a battery of toxicity tests are required. Because in such studies the experimental animals are often exposed to extremely high doses of the test compound, pathological changes may be detected which under more physiological conditions would not be observed. As an example, results of chronic and subchronic toxicity tests are presented in which laboratory animals have been fed on diets containing high concentrations of slowly digestible carbohydrates (lactose, raw potato starch, modified starches), sugar alcohols (xylitol, sorbitol, mannitol, lactitol) or bulking agents (polydextrose). Diarrhoea, caecal enlargement, pelvic nephrocalcinosis and adrenal medullary hyperplasia were noted in some of these studies as pathological end-points. Subsequent biochemical investigations revealed the underlying mechanisms for some of these effects. These data, together with results of clinical studies, indicated that the pathological effects observed in animal tests were of no significance for man, partly because of species differences and partly because of different levels of exposure. The results of these and other studies have drawn attention to the desirability of adopting more flexible approaches to the testing requirements for certain food additives, depending on such factors as their structure, occurrence in nature, and metabolism into normal metabolites, as well as their proposed use.