A study was undertaken to examine the potential of two non-genetically modified (non-GM) soy products, differentially processed, as fishmeal (FM) replacements. Dietary protein (DP) was replaced at 50, 60 and 70% using the 3011 meal or 40, 50 and 60% with the 3032 meal. Ingredient 3011 was a cooked and solvent-extracted meal from a high protein and low oligosaccharide soy cultivar; whereas, ingredient 3032 was a cold-pressed, low fat cake ground to a meal that originated from a medium protein, low oligosaccharide, low protease inhibitor soy cultivar. Experimental diets were isonitrogenous (45% crude protein, CP), isolipidic (13% lipid), isocaloric (12.6 kJ energy kg− 1 of diet) and were balanced for lysine (3.2%) and methionine + cysteine (2%). Taurine was supplemented at 1.5% in all diets. A commercial control diet (52% CP and 13% lipid) was included for a total of eight test diets. Twenty four round tanks (330 L) were each stocked with 20, 4 g yellowtail (Seriola lalandi). Tanks were supplied with recirculated seawater and each diet was randomly assigned to three replicate tanks. After a 10-week feeding trial, growth, survival, feed conversion ratio, protein efficiency ratio, protein retention, condition factor and carcass proximate composition were contrasted among dietary treatments. Results indicate that juvenile yellowtail fed diets in which the 3032 meal provided 50 and 60% of DP grew significantly (P < 0.001) better than those fed all other diets, including the FM-based (menhaden FM) and commercial control diets. Fish fed with the 3011 meal providing 50% of DP and those fed with the 3032 meal providing 40% of DP had similar growth performance which was slightly greater than the fish fed the FM reference diet. Minor intestinal histology differences were found among fish fed the 8 diets. Taken together these data indicates that the improved non-GM soybean meals supported better performance of juvenile yellowtail than fish fed the FM reference diet. No enteritis was detected even at high dietary inclusion levels.