Models of pig growth seek to predict the rates of retention of protein and lipid from a knowledge of the pig and the diet fed. The first, and crucial, part of the problem is to predict the rate of protein retention. A simple rule, which we have advocated, is that the marginal efficiency of retention of dietary ideal protein, ep, is the same at all liveweights for a given kind of pig, providing only that protein retention is below the potential of the pig. The objective of the experiment was to test this proposed rule by measuring ep in genetically similar pigs of different liveweights.