This paper describes an accurate and time-efficient method for the determination of total body potassium via a combination of measurements in the Birmingham whole body counter and the use of the Monte Carlo n-particle (MCNP) simulation code. In developing this method, MCNP has also been used to derive values for some components of the total measurement uncertainty which are difficult to quantify experimentally. A method is proposed for MCNP-assessed body habitus corrections based on a simple generic anthropomorphic model, scaled for individual height and weight. The use of this model increases patient comfort by reducing the need for comprehensive anthropomorphic measurements. The analysis shows that the total uncertainty in potassium weight determination by this whole body counting methodology for water-filled phantoms with a known amount of potassium is 2.7% (SD). The uncertainty in the method of body habitus correction (applicable also to phantom-based methods) is 1.5% (SD). It is concluded that this new strategy provides a sufficiently accurate model for routine clinical use.