Lab activities are a vital part of physics courses, providing students with hands-on experiences of classroom topics. At the same time, building self-efficacy is vital to student learning and retention; however, females traditionally have lower self-efficacy in STEM fields. This research explores the role of gender and the changes in interest in physics and the relation to changes in self-efficacy, as well as the interactions with math ability, the number of labs, and hours spent in the lab. Data from 260 institutions containing responses from over 11,000 women and 14,000 men was used. The self-efficacy construct was created from the E-CLASS survey. A two-level multi-level model was used to analyze the data. Level one variables included gender and change in physics interest. Level two variables included number of labs, hours spent in the lab, and math level; the dependent variable was change in self-efficacy. Change in interest positively correlated with change in self-efficacy, with women having a more favorable change. While the number of labs per semester was not a significant predictor of change in self-efficacy, females could tolerate more labs for no change in self-efficacy than men. Further, the gender effect (greater favorable change in self-efficacy for women) reverses with men showing a greater favorable change in self-efficacy than women after approximately 3.5 hours in the lab. Finally, math level had a positive correlation with change in interest and change in self-efficacy, with algebra-based labs demonstrating a more pronounced effect.