Objective To explore the correlation between plasma pre-treatment brain-derived neurotrophic factor （BDNF） and vascular endothelial growth factor （VEGF） levels， and curative effect in the acute stage as well as cognitive function in patients with first-episode major depressive disorder. Methods A total of 67 patients with first episode major depressive disorder admitted to the Affiliated Brain Hospital of Guangzhou Medical University from November 2016 to December 2019 were enrolled， and treated with escitalopram or venlafaxine for 4 weeks. The depressive symptoms and cognitive function before and after treatment were evaluated by Hamilton Depression Rating Scale （HAMD-17） and MATRICS Consensus Cognitive Battery （MCCB）. The levels of plasma BDNF and VEGF before treatment were detected. Results Compared with those before treatment， the scores of HAMD-17 decreased ［（22.3±4.7） points vs （8.5±5.6） points， t=19.649， P ＜ 0.001］， processing speed ［（43.7±10.8） points vs （47.8±10.7） points， t=-3.676， P ＜ 0.001］， working memory ［（40.5±11.1） points vs （44.8±11.2） points， t=-3.288， P=0.002］， the scores of verbal learning and memory function ［（39.2±10.2） vs （46.2±10.5）， t=-5.565， P ＜ 0.001］ increased， and the differences were statistically significant. BDNF and VEGF before treatment were negatively correlated with HAMD-17 （β= -0.265， P=0.036； β=-0.284， P=0.021）. BDNF was positively correlated with processing speed and working memory （β=0.305， P=0.018； β=0.416， P=0.001）. HAMD-17 showed statistically significant direct （β=2.602， P=0.013； β=1.036， P=0.021） and indirect （β=0.215， P=0.024； β=0.320， P=0.035） effects on the correlation between BDNF and speed of processing as well as working memory. BDNF before treatment was positively correlated with the changes in HAMD-17 （before treatment minus after treatment） at the end of 4 weeks treatment （β=2.009， P=0.030）， and was negatively correlated with changes in processing speed and working memory （β=-0.288， P=0.024； β=-0.267， P=0.039）. Changes in HAMD-17 scores showed statistically significantly indirect effect （β=-0.529， P=0.001； β=-0.523， P=0.002） between BDNF and changes in speed of processing and working memory， but no statistically significant direct effect （β=-0.236， P=0.321； β=-0.382， P=0.206）. Conclusions The level of plasma BDNF in patients with first episode depressive disorder before treatment can predict the antidepressant effect after 4 weeks of treatment， but it is not directly related to the improvement of cognitive function.