OBJECTIVE： To compare the effect and safety of agomelatine and paroxetine on cognitive function in first-episode depression patients. METHODS： 84 first-episode depression patients were randomly divided into agomelatine group（43 cases） and paroxetine group（41 cases）. Agomelatine group was given Agomelatine tablet with initial dose of 25 mg， increasing to 50 mg 2 weeks later， qd； paroxetine group was given paroxetine with initial dose of 20 mg， increasing to 50 mg 2 weeks later， qd. Treatment course of 2 groups lasted for 6 weeks. They were not given any other anti-depression agent. NCT， WMS-RC and MoCA score and the occurrence of ADR were observed in 2 groups before and after treatment. RESULTS： There was no statistical significance in NCT， WMS-RC and MoCA between 2 groups before treatment （P＞0.05）； after treatment， those of 2 groups improved significantly， and the rate of NCT error reduced significantly， with statistical significance （P＜0.05）. After treatment， the rate of NCT error in agomelatine group was lower than in paroxetine group， while the NCT， WMS-RC， MoCA scores and attention， recall and delayed recall score were higher than paroxetine group， with statistical significance （P＜0.05）. The incidence of ADR in agomelatine group was significantly lower than in paroxetine group， with statistical significance （P＜0.05）. CONCLUSIONS： Agomelatine is better than paroxetine in improving the first-episode depression patients in cognitive function and safety.