This study investigated effects of cognitive and motor dual task gait training on dual task gait performance in stroke. Participants (n = 28) were randomly assigned to cognitive dual task gait training (CDTT), motor dual task gait training (MDTT), or conventional physical therapy (CPT) group. Participants in CDTT or MDTT group practiced the cognitive or motor tasks respectively during walking. Participants in CPT group received strengthening, balance, and gait training. The intervention was 30 min/session, 3 sessions/week for 4 weeks. Three test conditions to evaluate the training effects were single walking, walking while performing cognitive task (serial subtraction), and walking while performing motor task (tray-carrying). Parameters included gait speed, dual task cost of gait speed (DTC-speed), cadence, stride time, and stride length. After CDTT, cognitive-motor dual task gait performance (stride length and DTC-speed) was improved (p = 0.021; p = 0.015). After MDTT, motor dual task gait performance (gait speed, stride length, and DTC-speed) was improved (p = 0.008; p = 0.008; p = 0.008 respectively). It seems that CDTT improved cognitive dual task gait performance and MDTT improved motor dual task gait performance although such improvements did not reach significant group difference. Therefore, different types of dual task gait training can be adopted to enhance different dual task gait performance in stroke.