In this paper I examine the anatomy of the price captive-supplies relationship to ascertain if some of the interpretations offered in the empirical literature are defensible. Gardner's one-product, two-input model is extended to consider a partially integrated oligopsonistic industry. The main result is that, although the empirical relationship between captive supplies and the price received by independent producers is negative, it may or may not be attributed to noncompetitive conduct. Hence, for an econometric model to detect what type of conduct the relationship reflects, more structural detail is needed than what so far has been provided in the literature. Copyright 1998, Oxford University Press.