Water is provided to the body in the form of obvious fluids, the water contained in solid foods, and the water produced in the oxidation of foodstuffs in the body; it is dissipated as vapor through the skin and lungs, and as liquid through the kidneys, bowel and skin. The common practice of attempting to estimate water balances by comparing the volume of the urine with the obvious fluids ingested is, therefore, quite inaccurate. The water of the food, drink, urine and feces may be determined directly. The water derived from the oxidation of foodstuffs and that lost through the skin and lungs can not be so simply measured. An indirect method for the estimation of these fractions of the water exchange has, therefore, been proposed by Newburgh (1). Estimations of water exchange by this method have yielded improbable results in many of our cases. Failure to obtain good results can not be attributed to disregard of the prescribed conditions, as will be shown below, nor is there reason to doubt the accuracy of the observations. Since this is so, a report of our experiences and a comprehensive investigation of the errors of the method seem warranted.It has been shown earlier that the calculations of Newburgh may be simplified by the use of the following formula, derived without additional assumptions (2)