It has been proposed that all animals have an inherent relationship between body protein and lipid that can be described allometrically, and the hypothesis tested in the research reported here is that the animal will at all times attempt to retain this relationship. The test was accomplished by feeding broilers, of three genotypes and in two experiments, in such a way as to produce lean and fat birds that were then subjected to a range of dietary protein levels in a second feeding period, during which their performance was measured. Birds were initially offered one of two feeds with widely different protein to energy ratios until they reached a pre-defined liveweight, after which they were given one of two feed protein contents in Experiment 1 and four in Experiment 2. Their performance was monitored until a second pre-defined liveweight was reached, at which time they were killed for carcass analysis. The genotype selected to be lean, in Experiment 1, showed no response to protein level in the second period, whether they were fat or lean at the start. Conversely, the genetically fat birds showed some additional growth in males and additional efficiency in the females. Averaged across genotypes and sexes, birds initially in the fat state gained only 6.9 g lipid/d versus 13.5 g lipid/d for the nutritionally lean broilers. In Experiment 2, growth rate and feed conversion efficiency (FCE) were related directly to dietary protein content and were higher for those birds made nutritionally fat. Carcass lipid gain was lower for the initially fat birds on the three highest dietary protein treatments. All birds made fat at 880 g and 1000 g, by giving them a low protein feed, had a much reduced fat content in their subsequent gain, provided that the protein content of the feed used was sufficiently high, indicating that they were making use of the excessive lipid reserves as an energy source. The hypothesis tested cannot be rejected by the evidence presented.