The 1970 geographical distributions of total anesthesia manpower, anesthesiologists, and nurse anesthetists by state were analyzed by multiple regression to account for their unevenness. Independent variables included factors relating to prior professional contact in the state, professional satisfaction, practice income, demand for services, and environment. The distribution of training programs accounted for 41 per cent of the variance in the distribution of total manpower, but 55 per cent was explained by the number of operations, location of nurse anesthesia schools, and proportion of total state employment in service occupations (a proxy for the availability of consumer services). Location of training programs and the absence of the other type of personnel were good predictors for the manpower subgroups. The distribution of nurse anesthesia schools, anesthesiologists, number of surgical operations, and the relative value schedule conversion factor together accounted for 60 per cent of the variance in the nurses' distribution. The location of residency programs (or positions) was a better predictor for the anesthesiologists') location than medical schools or factors characterizing the demand for services. The distribution of nurse anesthetists, hospital cost per day (considered a proxy for a satisfying professional life and for regionalization of services), and residency programs explained S1 per cent of the variance in the anesthesiologists' distribution. Although the regression predicts that increasing the number of residency programs in an underserved state should be associated with an increase the number of anesthesiologists, such a policy may be infeasible due to pending federal health manpower legislation unless matched by decreasing a greater number of programs in relatively oversupplied states.