People frequently encounter advice when making decisions in daily life. Advice is critical to improving decision quality. However, decision-makers are not such receptive to take in advice due to various situational and dispositional factors. Cognitive confidence has been confirmed as one of the most important factors that hinder advice taking, as confident decision-makers are more likely to place more weight on their own initial judgment and disregard others’. Yet, little is known about the impact of dispositional confidence, namely self-confidence on advice taking. Self-confidence, which comprises both explicit and implicit components resting with the degree of consciousness, is defined as one’s beliefs in his/her ability to handle over something (e.g., task at hand). Thus, from the self-concept maintaining perspective, decision-makers with a high level of self-confidence would be less likely to take advice for maintaining a competent self-view. Following the same logic, task complexity is supposed to be a situational factor that affects how receptive decision-makers are to advices. Specifically, effects of self-concept maintaining motivation would be enhanced as task complexity increases as the achievement of a complex task is a more credible competence certificate. On the other hand, explicit and implicit self-confidence discrepancies are supposed to impair self-concept maintaining motivation for implicit self-doubt resulted from discrepancies. In this vein, self-concept maintaining motivation would be further weakened in complex task settings and decision-makers would be more likely to share the extra responsibilities and risks with advisors by taking advice. In the current study, we therefore aim to investigate the joint effect of explicit and implicit self-confidence on advice taking, and the moderating role of task complexity. Two experiments were conducted to testify the foregoing hypotheses. Explicit and implicit self-confidence were measured by a self-report questionnaire and the Single Category Implicit Association Test respectively. Participants completed the coin quantity estimation task, while the task complexity of study 2 was increased by introducing an incentive system. Participants engaged in a repeated estimation task after the initial estimation but at the second time they received advice from a fictitious advisor. Compared with the free advice in study 1, advice in study 2 must be paid before accessing. We measured the extent to which participants relied upon advice using the weight of advice (WOA) index. In study 1, one hundred and twenty-three (under)graduate students were recruited. The results showed that explicit self-confidence was negatively related to advice taking, implicit self-confidence, self-confidence discrepancies, and direction of discrepancies had no impact on advice taking in simple task settings. One hundred and ten (under)graduate students participated in study 2. The results showed that explicit self-confidence was negatively and self-confidence discrepancies were positively related to advice taking, implicit self-confidence and direction of discrepancies had no impact on advice taking in complex task settings. Results based on multilevel analysis confirmed that task complexity amplified the effects of self-confidence discrepancies on advice taking (regression coefficient changing from 0.08 to 0.33). Implications, limitations and future directions are discussed.