Sixty Merino lambs weighing 25 kg were randomly allotted to six different feeding levels as follows: ad libitum, 80%, 71%, 65%, 55% and 45% ad libitum. Individual intakes and live masses were determined weekly. From 25 – 33 kg live mass the lambs received restricted feeding except for the control groups on ad lib intakes. At 33 kg live mass, half of each group was slaughtered and the remainder received ad libitum feed to 45 kg live mass. The 55% and 45% ad libitum groups were put on ad libitum feed at 31 and 28 kg live mass because the ad libitum groups had already reached 45 kg live mass and had been slaughtered. During the restriction phase, efficiency of feed conversion decreased as restriction increased. For the first week of the realimentation phase, efficiency of feed conversion increased as restriction increased. From the second week of the realimentation phase, efficiency of feed conversion was the same for the ad libitum, 80%, 71% and 65% ad libitum groups (79,4 MJ DE/kg LWG) while the efficiency of feed conversion for the 55% and 45% ad libitum groups decreased to 86,1 and 94,1 MJ DE/kg LWG, respectively.