Abstract Due to the use of various, and mostly indirect, methods to estimate total body water (TBW) and extracellular water (ECW), there is no agreement about whether body water distribution, i.e. the ECW to TBW ratio, is normal in GH-deficient (GHD) subjects at baseline and during recombinant human GH (rhGH) treatment. We studied body water distribution in 14 patients with adult-onset GHD and in 28 healthy controls. We also investigated the effect of GH replacement therapy for 4 and 52 weeks on body water distribution. All patients started with a dose of 0.6 IU rhGH/day for the first 4 weeks. After 52 weeks, the dose varied between 0.6–1.8 IU/day. TBW and ECW were measured by dilution of deuterium and bromide, respectively. Both parameters were also estimated using multifrequency bioelectrical impedance (BIA). Patients with GHD had significantly lower ECW and TBW than healthy controls. In addition, the ECW to TBW ratio was significantly lower in GHD patients than in healthy controls. Four weeks of GH treatment significantly increased body weight, TBW, ECW, and ECW/TBW. A further increase in TBW, but not ECW, was found after 52 weeks of treatment. The mean increases in TBW and ECW from the baselines were 2.5 ± 0.3 and 2.0 ± 0.3 L, respectively. The correlation coefficient and the estimated reliability between measured and estimated TBW and ECW at any time point were all high (>0.91 and >0.95, respectively). In general, both ECW and TBW were overestimated by multifrequency BIA in GHD adults. During treatment, the overestimation of both ECW and TBW diminished. The estimation error was correlated with the level of the body water compartment and the ratio of ECW to TBW. The estimated change in ECW with rhGH treatment was underestimated by multifrequency BIA. We conclude that GHD adults have lower ECW and TBW and a lower ECW to TBW ratio, as measured by dilution techniques. The ECW to TBW ratio can be normalized within 4 weeks of rhGH treatment at a dose of 0.6 IU/day. Finally, we conclude that multifrequency impedance measurements do not give valid estimates of body water compartments in the follow-up of patients with GHD.