Abstract: To optimize the preparing process of cobalt powders by hydrogen reduction of cobalt oxide and improve the product quality, the single-factor experiments were first conducted to identify the key influencing factors as loading capacity, reduction temperature, and reduction duration. Subsequently, the Box-Behnken design of response surface methodology was adopted to investigate the process conditions, and the prediction model for cobalt powder quality was established and verified. The results show that the model exhibits the excellent fitting performance with determination coefficient (R²=0.9918). The predicted values are in good agreement with the measured values, enabling the accurate prediction of cobalt powder quality. The optimal process parameters are determined as follows: boat loading capacity of 68 g, reduction temperature of 398 ℃, and reduction duration of 186 min. Under these conditions, the cobalt powders have the particle size of 0.98 μm, oxygen mass fraction of 0.380%, carbon mass fraction of 0.011%, and span of 1.54 with the comprehensive quality score of 83, presenting the optimal overall performance. The variance analysis indicates that the influencing significance of the factors follows the order as reduction temperature>reduction time>boat loading capacity, and the interactive effect exists among these factors.