Standard techniques conventionally used to assess body composition have various drawbacks which include a requirement for specialized equipment and expertise not widely available, radiation exposure, extensive study time periods, and limited clinical utility. Anthropometric methods, which are more clinically feasible, also involve substantially greater calculation error. This study was designed to compare a newly developed whole-body magnetic resonance (MR) imaging technique with standard tritium dilution (THO) to measure total body water (TBW) in eight primates (Macaca fascicularis). Sedated primates underwent whole-body MR imaging (10-12 contiguous 50-mm slices) and TBW was computed using a previously validated program. After 5 days to allow for primate recovery and stabilization, TBW measurements were repeated using THO methodology. Linear regression analysis of TBWMRI (72.1 ± 5.7% body wt) compared with TBWTHO (73.8 ± 3.2% body wt) yielded a Pearson correlation of r = 0.8145, P = 0.02 with a mean squared error of 2.14. Whole-body MR imaging determination of TBW correlates well with THO, requires less time (20 min versus 24 hr), and does not expose the subject to radiation. This study suggests that whole-body MR imaging is an accurate technique for in vivo body composition analysis. As demonstrated by the body size of this investigational primate model, the technique may be particularly applicable to the study of human infants. © 1992.