Genetical and economical consequences of different breeding programs for laying hens were simulated with percent residual food consumption (PRFC) included in multi-trait selection indices. The breeding program was calculated to be 17% more efficient when PRFC was included in the sire and dam indices based on all available sources of information, compared with no information on food consumption. Records of food consumption on hens only are expected to be somewhat cheaper and almost as efficient as recording on both sexes (14%). A less expensive method, but one also calculated to be significantly less efficient (4%), is to record PRFC exclusively on cocks. The efficiency of the breeding program was calculated only to increase by 2% when preselecting for food consumption in very young chicks, and the correlated selection response in PRFC was much lower than the other alternatives discussed.