A regression analysis was conducted of factors that help to explain the variance in physician distribution and which identify those factors that influence the maldistribution of physicians. Models were developed for different geographic areas to determine the most appropriate unit of analysis for the Western Missouri Area Health Education Center region. The data source was the area resource file (ARF) originally developed by the U.S. Bureau of Health Manpower. The subprogram REGRESSION from the Statistical Package for the Social Sciences was used in the analysis. Four models were developed comparing counties and Rand McNally Minor Trade Areas for the entire United States and for the four states comprising Federal Region VII. The four models explaining (or predicting) physician to population ratio allow for a cross-comparison between the county and trade areas unit of analysis at the regional and national level. It was found that more densely settled areas have more physicians and hospital beds but the ratio of physicians to population and hospital beds to population decreases with greater density. It is suggested that residual analysis based on trade areas would probably be more applicable to Missouri. A refined causal analysis model(s) can be of assistance to policy-makers, evaluators, and planners. A review of previous research is included. (SW)