Predicting worker’s effort is important in many different areas, but is often difficult. Using a laboratory experiment, we test the hypothesis that confidence, i.e. person-specific beliefs about her abilities, can be used as a generic proxy to predict effort provision. We measure confidence in the domain of financial knowledge in three different ways (self-assessed knowledge, probability-based confidence, and incentive-compatible confidence) and find a positive relation with the actual provision of effort in an unrelated domain. Additional analysis shows that the findings are independent of personal traits such as gender, age, and nationality.