Accurate energy values for diet formulation help meet broiler performance and economic objectives. Previous research has shown that energy balance of test ingredients may be affected by assay methodology. Two experiments were conducted to evaluate three assay methodologies to determine digestible (DE) and metabolizable (ME) energy of corn and soybean meal. Experiment 1 evaluated corn and consisted of six dietary treatments, including a semi-purified diet, a direct diet (94% corn meal), two basal diets, and two diets where corn was substituted at either 15 or 30% into respective basal diets. Experiment 2 evaluated soybean meal and consisted of six dietary treatments, including a semi-purified diet, a direct diet (43% soybean meal + 52% dextrose), two basal diets, and two diets where soybean meal was substituted at either 10 or 20% into respective basal diets. Basal diets in both experiments were formulated to provide 1.15% digestible Lys in the substitution diets. The semi-purified diet was included to estimate endogenous losses of energy. All diets contained titanium dioxide as an indigestible marker. In both experiments, day-old YP × Ross 708 male chicks were placed into battery cages (9 chicks/cage) and were provided a common starter diet from 1 to 17 D of age. At 18 D of age, birds were provided with test diets. Excreta was collected at 21 and 22 D of age and feed was weighed at 22 D of age. Terminal ileal digesta was collected at 23 and 24 D of age. Titanium dioxide concentration, N, and gross energy were determined in excreta and ileal digesta samples. Each experiment was conducted as a randomized complete block design with 16 replicates per diet. Apparent DE and ME, standardized DE and ME, and N-corrected apparent ME were determined for both corn and soybean meal using the direct method and the substitution method at two inclusions. Energy utilization was analyzed using analysis of variance. In Experiment 1, determined apparent and standardized ME of corn were reduced utilizing the direct method compared with substitution at either 15 or 30% (P < 0.0001; 3,290, 3,635, or 3,636 kcal/kg AME, respectively). In Experiment 2, determined apparent and standardized ME and N-corrected ME of soybean meal were reduced in the direct method compared with substitution at 10 or 20% (P ≤ 0.0024; 2,211, 2,496, or 2,437 kcal/kg AME, respectively). In both experiments, standard error was increased using the substitution method compared with the direct method due to attributing all variance in dietary energy to the test ingredient.