Currently, energy evaluation of fish feeds is performed on a digestible energy basis. In contrast to net energy (NE) evaluation systems, digestible energy evaluation systems do not differentiate between the different types of digested nutrients regarding their potential for growth. The aim was to develop an NE evaluation for fish by estimating the energy efficiency of digestible nutrients (protein, fat and carbohydrates) and to assess whether these efficiencies differed between Nile tilapia and rainbow trout. Two data sets were constructed. The tilapia and rainbow data set contained, respectively, eight and nine experiments in which the digestibility of protein, fat and energy and the complete energy balances for twenty-three and forty-five diets was measured. The digestible protein (dCP), digestible fat (dFat) and digestible carbohydrate intakes (dCarb) were calculated. By multiple regression analysis, retained energy (RE) was related to dCP, dFat and dCarb. In tilapia, all digestible nutrients were linearly related to RE (P<0·001). In trout, RE was quadratically related to dCarb (P<0·01) and linearly to dCP and dFat (P<0·001). The NE formula was NE=11·5×dCP+35·8×dFAT+11·3×dCarb for tilapia and NE=13·5×dCP+33·0×dFAT+34·0×dCarb–3·64×(dCarb)2 for trout (NE in kJ/(kg0·8×d); dCP, dFat and dCarb in g/(kg0·8×d)). In tilapia, the energetic efficiency of dCP, dFat and dCarb was 49, 91 and 66 %, respectively, showing large similarity with pigs. Tilapia and trout had similar energy efficiencies of dCP (49 v. 57 %) and dFat (91 v. 84 %), but differed regarding dCarb.