Hydrostatic weighing and chemical dilution are well accepted methods for measuring body composition. Recently, Dual Energy X-ray Absorptiometry (DEXA) has become the preferred method. The two compartment algorithms used by these methods assume a fixed constant for lean body tissue. This constant has long been suspect of variations due to many reasons such as ethnic or racial differences, age, health, etc. Variations in muscle mass, however, are factors that affect every single subject. The study presented here includes muscle mass as one compartment of a three compartment body composition model. Each compartment is modeled as a function of body water measurements with hydrometric constants as parameters. A muscle mass algorithm is hypothesized and tested to support the authors' contention that muscle mass can be approximated by making body water measurements. It is demonstrated that, when muscle mass becomes a factor, the traditional two compartment models can overstate body fat ratios by as much as four percent. (Contains 2 tables and 3 figures.)