Selection based on individual feed conversion ratio was made for three generations in a chicken population previously selected for rate of growth and for one generation in a randombred population which had not been selected for rate of growth. Two sets of high (HL) and two sets of low (LL) feed conversion ratio lines were selected in each population: one set selected so that body weight was constant over generations (HLWK, LLWK), while in the other set (HLWV, LLWV) body weight was allowed to vary.In the three generations of selection in the chicken population the units of feed required for a unit of gain was increased .12 in the HLWV line over that of non-selected controls and .08 in the HLWK line. In the low feed conversion ratio lines the units of feed required for a unit of gain was decreased .11 in the LLWK line and .07 in the LLWV line. In both low lines more progress was made in females than in males. In the randombred stock, which had no previous selection for growth, as much progress in lowering the feed conversion ratio was made in one generation of selection as was made in three generations in the broiler population (decreased .12 in the LLWK and .13 in the LLWV) with no sex difference in response.The realized heritabilities for feed conversion for the lines selected from the broiler population were .20, .21, .33 and .26 for the LLWV, LLWK, HLWV, HLWK, respectively. In the lines selected from the randombred population the realized heritabilities were .56, .45, .35 and .31, respectively.The magnitude of the heritability values obtained indicate that even in lines previously selected for growth there is sufficient variation in efficiency of feed utilization to allow progress in selection without further change in body weight.