Live-weight gain, food intake and growth of chemical components of the empty body were determined in 32 pigs (16 of each sex) growing between 6 and 30 kg live wieght and offered diets based on cooked cereals and fat-filled skim milk with either soya-bean meal (SBM) or herring meal (HM) as the source of supplementary protein. The diets were given ad libitum or to a restricted scale and with or without Avotan (40 mg avoparcin kg-1). Diets with HM gave a higher dry matter (DM) content of the empty body than SBM (341.2 vs. 323.5 g kg-1; P < 0.001), higher fat and energy contents (359.4 vs. 322.0 g kg-1 DM; 27.06 vs. 26.10 MJ kg-1 DM; P < 0.01) and lower protein content (531.5 vs. 558.5 g kg-1 DM; P < 0.05). Ad libitum feeding resulted in a higher DM content of the empty body than scale feeding (338.8 vs. 325.9 g kg-1; P < 0.05) and a higher fat content (368.0 vs. 313.4 g kg-1 DM; P < 0.001). The use of HM gave greater levels of fat gain in the empty body than did SBM (62.5 vs. 48.8 g per day; P < 0.01) and a greater efficiency of energy retention (0.386 vs. 0.339; P < 0.001) and nitrogen retention (0.633 vs. 0.581; P < 0.01). The two protein sources gave similar rates of gain of live weight (515 vs. 509 g day-1) and of protein in the empty body (91.3 v. 88.6 g per day). The males and females had similar gains except that the gain of water in the empty body was higher for the males (332.3 vs. 311.5 g per day; P < 0.05). The males had a better food conversion ratio than the females (1.269 vs. 1.307; P < 0.05). Pigs fed ad libitum grew faster than those fed to the restricted scale (573.4 vs. 451.0 g per day; P < 0.001) and had a greater gain of chemical components of the empty body and greater efficiency of energy retention. The two groups had similar food conversion ratios. Inclusion of Avotan had little effect on any of the parameters, except that the food conversion ratio was improved (P < 0.05) when the diet contained both HM and Avotan.