To investigate the correlation between brain-derived neurotrophic factor (BDNF) and risk factors, as well as functional outcome in poststroke depression (PSD) or poststroke anxiety (PSA). Cohort study. Stroke patients admitted to an urban rehabilitation hospital. Stroke patients (N=162) without any previous history of depression and anxiety. Not applicable. Sociodemographic information and comorbidities were recorded during hospital admission. Functional outcomes were assessed using FIM scores at time of admission and discharge. The influence of various factors such as BDNF and patient characteristics on functional outcome was investigated. Single-factor effect was examined using simple logistic regression, as was multi-factor effect using multiple logistic regression. The goodness-of-fit of those regression models was evaluated by the integrated area under ROC curve. PSD was diagnosed in 61 (37.7%) patients, and PSA was diagnosed in 40 (24.7%). Multiple logistic analysis showed that BDNF, divorce or separation, and history of smoking were significantly associated with the occurrence of PSD but not with the occurrence of PSA. The model combining low BDNF level and divorce or separation improved the prediction for PSD. Among the variables analyzed for prediction of functional outcome, serum BDNF had a minimum correlation with motor FIM scores in PSD but no significant correlation with motor FIM scores in PSA. BDNF is a valuable prediction for the occurrence of PSD but not for PSA. More strikingly, ischemic stroke patients who are divorced or separated with low serum BDNF have a much higher risk for PSD. BDNF has a minimum correlation with motor function outcome in PSD but no significant correlation with motor outcome in PSA. Copyright © 2020 American Congress of Rehabilitation Medicine. Published by Elsevier Inc. All rights reserved.