A 10-week feeding trial was conducted to estimate the optimum dietary manganese requirement for juvenile cobia, Rachycentron canadum L. The basal diet was formulated to contain 501 g kg−1 crude protein from vitamin-free casein, gelatin and fish protein concentrate. Manganese sulphate was added to the basal diet at 0 (control group), 6, 12, 18, 24 and 36 mg Mn kg−1 diet providing 5.98, 7.23, 16.05, 23.87, 28.87 and 41.29 mg Mn kg−1 diet, respectively. Each diet was randomly fed to three replicate groups of cobia for 10 weeks, and each tank was stocked with 30 fish (initial weight, 6.27 ± 0.03 g). The manganese concentration in rearing water was monitored during the feeding period and was < 0.01 mg L−1. Dietary manganese level significantly influenced survival ratio (SR), specific growth ratio (SGR), feed efficiency ratio (FER) and the manganese concentrations in the whole body, vertebra and liver of cobia. When the dietary manganese level rose from 5.98 mg kg−1 to 23.87 mg kg−1, the superoxide dismutase (SOD; EC 22.214.171.124) activities in liver also increased (P < 0.05). But there was no significant change in SOD activities for the groups fed with diets containing manganese level higher than 23.87 mg kg−1. On the basis of broken-line regression of SGR, manganese concentration in whole body and vertebra the manganese requirements of juvenile cobia were 21.72 mg kg−1, 22.38 mg kg−1 and 24.93 mg kg−1 diet in the form of manganese sulphate, respectively.