The aim of this study was to determine the kinetics of H(2)18O equilibration and elimination in preterm babies for the estimation of total body water. Thirteen, clinically stable, preterm babies of less than 32 weeks' gestation were studied in the first week after birth. Blood and urine samples were obtained for baseline measurement of 18O:16O ratio and 1 ml/kg of 10% H(2)18O (0.1 g/kg isotope) administered orally. Eleven blood samples were obtained over the next six hours and between one and four over the next 18 hours. During the same 24 hour period between three and eight urine samples were also obtained. The dilution space at zero time (volume of distribution or total body water) was estimated using double exponential curve fitting using all available points, from single samples and from linear regression on the log data using two or three samples. Equilibration time was variable and showed a significant correlation with percentage change in body weight from birth. For blood samples, the median time to equilibrium was 81 minutes (range 2 to 191). A plateau phase was not detected, with H(2)18O enrichment declining after the point of maximum enrichment. The median volume of distribution at time 0, based on double exponential curve fit analysis, was 859 ml/kg (range 755 to 995). The volume of distribution, estimated from linear regression on the log data using two serum samples obtained at three and six hours, approximated most closely to that based on exponential curve fit analysis with a median difference of -4 ml/kg (range -41 to 73). It was concluded that in most situations blood sampling at three and six hours may be acceptable. However, as equilibration time is variable and influenced by the state of expansion or depletion of body water compartments, when studying overhydration states, multiple sampling is advised in order to be certain that the elimination phase has been reached.