We had the opportunity of investigating a case (BK) of a severe form of mucopolysaccharidosis with nearly total deficiency of beta-glucuronidase in serum, leucocytes and fibroblasts. We here report results obtained by prenatal diagnosis of a clinically normal child (BK's sister), and point out the difficulty in interpreting a heterozygous level of beta-glucuronidase activity in cultured amniotic cells. Four successive passages of amniotic cells were tested for beta-glucuronidase and alpha-mannosidase activity in at-risk and control cells. In different passages, enzyme activity was between 8 and 49 per cent of controls but 2 to 18 times higher than fibroblasts from the affected brother (BK). The highest activity was observed in the first passage and the lowest in the third. The electrophoretic separation of GAGS from at-risk amniotic fluid showed a normal pattern. We discuss the correlation between enzyme levels in different passages of cultured cells and that found in leucocytes and fibroblasts from the propositus and parents. From a practical point of view, we conclude that the first passage gives the most reliable results for prenatal diagnosis.