Reliability of differential sugar absorption tests is hampered by a lack of standardization of the content and osmolarity of the test solutions. We evaluated the effect of osmolarity of the test solution of the sugar absorption test on the 5 hour urine excretion of orally administered lactulose and mannitol. A group of 28 controls and 14 coeliacs, with villous atrophy grade II to IV, ingested a hyperosmolar sugar absorption test solution and a "low"-osmolar solution, respectively. After an overnight fast, each subject ingested hyperosmolar sugar absorption test solution (2 g mannitol, 5 g lactulose and 40 g sucrose/100 ml (around 1,560 mmol/l)). After two days, this procedure was repeated with low-osmolar solution (2 g mannitol and 5 g lactulose/100 ml (around 375 mmol/l). The influence of the sequence of the tests on the results had previously been excluded. All urine from the 5 h-period following ingestion of the test solution was collected. To calculate the low-osmolar solution ratio, samples were analysed for lactulose and mannitol concentrations by gas chromatography The sensitivity of hyperosmolar SAT solution and low-osmolar solution for the detection of mucosal abnormalities in coeliacs was 64% and 43%, respectively. In conclusion, a hyperosmolar solution discriminates better between normal and damaged mucosa of the small bowel such as villous atrophy due to a relative increase in permeability for lactulose.