Soybean oil (SO), an important plant-derived lipid, is routinely used in aquafeed as a partial substitute for fish oil. However, the effect of various contents of SO in a SO-based aquafeed on specific aquatic animals and the underlying mechanisms are unclear. In this study, five isonitrogenous diets containing different SO contents (0 %, 3 %, 6 %, 9 %, and 12 %, named as S0, S3, S6, S9, and S12, respectively) were prepared with SO as the predominant lipid source, and 12-week feeding experiments were conducted on juvenile grass carp (Ctenopharyngodon idella) (initial weight: 12.54 ± 0.08 g). Apart from lower feed conversion ratio in the S6 and S9 groups (P < 0.05), no obvious changes were observed in other growth indices (e.g., final weight and weight gain rate) in grass carp fed on diets with 0–9 % SO (P > 0.05). The S12 group exhibited significant decrease in the final weight and weight gain rate and increase in the viscerosomatic index (P < 0.05). Unlike gradual decrease in moisture, ash, and saturated fatty acid contents, muscle crude fat and unsaturated fatty acid (particularly polyunsaturated fatty acid) contents significantly increased with the increasing SO contents (P < 0.05). Serum triglyceride contents continuously increased with the increasing SO amounts, exhibiting the maximum value in the S12 group (P < 0.05). Similarly, the S12 group displayed the highest levels of aspartate aminotransferase, alanine transaminase, low-density lipoprotein cholesterol, and total cholesterol in serum (P < 0.05). Only the S12 group exhibited scanty lymphocytic infiltration in some areas of the hepatopancreas. However, other groups exhibited no obvious histopathological abnormalities. Lipogenic genes (dgat, fad, elovl1, elovl6, elovl7, lpcat, and pla2), ACSL family members (acsl1a, acsl1b, acsl3, acsl4, acsl5, and acsl6), lipolytic genes (atg1, cpt1a, cpt1b, hsl, and lpl), fatty acid transport-related genes (cd36, fabp, and fatp), and transcription factor involved in lipid metabolism (ppar α) were gradually upregulated as dietary SO content increased (P < 0.05). However, the transcript levels of srebp1 and its target gene acc1 exhibited decreasing trend under the same conditions, and significant downregulation of ppar γ was observed in the S12 group (P < 0.05). In conclusion, the optimal content of SO as the main lipid source in a SO-based diet for juvenile grass carp ranged from 6 % to 9 %, at which growth requirements of grass carp were met and fish health and fillet quality were effectively maintained. Our data provided a scientific reference for developing more efficient aquafeed formulations and enhancing fish flesh quality in aquaculture, as well as provided basic information for further assessing the regulatory mechanism of lipid homeostasis in teleost fishes.