Measurement of energy expenditure with doubly-labelled water and of body composition and breast milk output with 2H or 18O requires accurate and precise techniques for measuring isotopic enrichments. The possibility of an inaccuracy in measurements of 2H and 18O isotopic enrichment arising from the matrix in biological fluids was investigated (1) by simulating a dilution experiment in both water and urine samples and (2) by reconstituting urine samples, ranging from 10 to 60 g/kg in solid concentration, from freeze-dried urinary solids mixed with either natural abundance or doubly-labelled water. Current techniques involved in measuring 2H and 18O isotopic enrichments were used (reduction of the samples to H2 gas with either Zn or U, and CO2/H2O equilibration or direct measurement of mass 20:18 ratios on water vapour for 18O analysis). All four methods accurately measured serial dilutions in both urine and water. Dilution space calculated from isotopic enrichments, compared with the water content of urine (determined by freeze-drying and accounting for exchangeable isotopes) was overestimated by about 2.4 % by the Zn technique whereas other methods were accurate. The urinary solids content of a water solution was related to that inaccuracy. The use of the Zn technique with biological samples is likely to create biases in 2H distribution space. Examination of recent literature supports this view. Caution should therefore be used when physiological conclusions have to be made from the relative size of 2H and 18O distribution spaces.