From the perspective of knowledge transfer, this research makes up for the weaknesses made by Bos et al.(2008) who ignored the effect of internal factors on unconscious thought. This study suggests that the unconscious thought is both unconscious thought knowledge-dependent and external goal-dependent. There are two experiments to demonstrate this. Experiment 1 investigated whether unconscious thought is knowledge-dependent. 96 participants were assigned to two groups randomly, a group with knowledge-learning (participants who need to learn Dijksterhuis’s unconscious thought theory) and a group without knowledge-learning. Then, all participants should complete a car decision task in one of three thought conditions: immediate decision condition, conscious thought condition, and unconscious thought condition. We compared the different scores between conditions using a 2(knowledge-dependence: yes, no)×3(thought conditions: immediate decision, conscious thought, unconscious thought) analysis of variance.The results showed that (1) the interaction between knowledge-dependence and thought conditions was significant, F(2,84)=3.44, p< .05. (2) A simple effect of knowledge-dependence in unconscious thought condition (F(1,84)=11.79, p= .001) and conscious thought condition (F(1,84)=5.38, p< .05) was found significantly. In the condition of unconscious thought, participants with knowledge-learning (M=15.80, SD=2.18) performed better than participants without knowledge-learning (M=8.40, SD=5.37, p< .05); in the condition of conscious thought, participants with knowledge-learning (M=9.87, SD=5.93) also performed better than participants without knowledge-learning (M=4.87, SD=9.35), p< .05. (3)A simple effect of thought conditions in knowledge-learning(F(2,84)=11.03, p= .000) was found significantly. In the condition of knowledge-learning, participants who thought unconsciously(M=15.80, SD=2.18) performed better than participants who thought consciously(M=9.87, SD=5.93, p< .05) and participants who decided immediately(M=5.73, SD=5.85, p< .001), and the latter two did not differ significantly(p> .05). Experiment 2 investigated whether unconscious thought was double-directional in both knowledge and goal. 60 participants were assigned to a group with knowledge-learning and a group without knowledge-learning at random, then they were asked to achieve a car-roommate decison task. In this task, participants were either told that they would later be asked questions about the cars or they would later be asked questions about the roommates. At last, all particaipants should rate the cars and roommates. We compared the different scores between conditions using a 2(knowledge-dependence: yes, no)×2(goal-dependence: yes, no) analysis of variance. The results showed that (1) the interaction between knowledge-dependence and goal-dependence was significant, F(1,58)=5.92, p< .05. (2) the simple effect of knowledge-dependence in goal-dependence condition was significant, F(1,58)=10.03, p< .01. When they had a task goal, participants who learned unconscious thought theory (M=12.53, SD=5.36) performed better than participants who didn't learn (M=8.33, SD=4.09), p< .01. (3) the simple effect of goal-dependence in knowledge-learning condition was significant, F(1,58)=9.79, p< .01. In the condition of knowledge-learning, participants who had the goal to think unconsciously (M=12.53, SD=5.36) performed better than participants who didn't have the goal (M=8.63, SD=4.09), p< .01. All results of two experiments have demonstrated that unconscious thought is dependent on both unconscious thought knowledge and goal.