The metabolism of erythritol was assessed in six normal volunteers by measuring the amount of 13C02 excretion and H, excretion in breath, and erythritol in urine after intake of 25 g '3C-labelled erythritol. The results were compared with the same variables obtained after intake of 25 g I3C-labelled glucose and 13C-labelled lactitol. In addition, the H, production by faecal flora supplemented with small amounts of erythritol, glucose and lactitol was measured in vitro, as an index of bacterial metabolism of nonabsorbed substrate. In contrast to the results obtained after intake of glucose and lactitol, no increase in breath I3CO, and H, was observed after intake of erythritol, and erythritol was nearly completely recovered in urine. The in vitro experiments showed that no H, was formed by faecal flora from erythritol as compared with glucose and lactitol. It is concluded that erythritol is a substrate that is readily absorbed, and undergoes no metabolism by the host. If part of it escapes absorption, it is not metabolized by faecal flora.Erythritol metabolism: Glucose: Lactitol: Humans Recently various new sugars and sugar-alcohols have become available for nutritional and therapeutic use. They are proposed as low-energy sweeteners, less cariogenic than glucose or sucrose, but, in contrast to artificial sweeteners such as aspartame and saccharin, they provide a bulking effect to foodstuffs. Some of them, e.g. lactitol and lactulose, are also used in the treatment of constipation and portal-systemic encephalopathy. Erythritol is a C, polyol, naturally occurring in algae and mushrooms, with a pleasant taste and interesting physico-chemical characteristics (Oku & Noda, 1990). The aim of the present study was to determine the metabolic characteristics of erythritol in humans.The evaluation of the energy value of a new product is cumbersome. Feeding studies can be used in animals but are difficult to perform in man (Van Es, 1987). Another technique of evaluation is based on the measurement of the fate of a dose of radiolabelled substrate (Grimble et al. 1988). Both techniques require a number of assumptions, and the energy value obtained should be considered as approximate. In the present study we evaluated the metabolic characteristics of erythritol by comparing it with glucose and lactitol, two substrates with a well-identified metabolism, by means of the following tests: 13C0, breath tests, H, breath tests, and measurement of urinary substrate recovery. Because colonic fermentation is an important mechanism of energy salvage for non-absorbed carbohydrates, we have also compared the amount of H, formed by colonic flora in a faecal incubation system to which the different substrates were added.