Background: Completing long-term antituberculosis treatment remains a challenge in Indonesia. Self-efficacy, a patient's confidence in his or her ability to complete treatment, is considered an important resource for adherence. Tuberculosis (TB) knowledge and the health beliefs described by the Health Belief Model (HBM) are theoretically related to self-efficacy, but evidence on how they are jointly associated with self-efficacy among patients treated in Indonesian primary care is limited. Objective: To examine the associations of TB knowledge and health beliefs with self-efficacy among patients receiving pulmonary TB treatment. Methods: This cross-sectional study included 72 patients with pulmonary TB recruited consecutively at the TB/DOTS clinic of Kalijati Community Health Center, Subang Regency, West Java, from June to August 2026. Knowledge, health beliefs, and self-efficacy were measured with a 13-item knowledge questionnaire, a 20-item HBM questionnaire, and the 25-item TB Self-Efficacy Scale (TBSE). Data were analyzed using Spearman's rank correlation and multivariable binary logistic regression. Results: Most participants had moderate self-efficacy (75.0%), and none had low self-efficacy. Knowledge showed a weak inverse correlation with self-efficacy (rs = −0.275; p = 0.019), whereas the total HBM score showed a positive correlation (rs = 0.394; p = 0.001). In the multivariable model (outcome: high vs. moderate self-efficacy), each one-point increase in the HBM score was associated with higher odds of high self-efficacy (adjusted odds ratio [AOR] 4.00; 95% CI 1.69–9.50), each one-point increase in the knowledge score was associated with lower odds (AOR 0.32; 95% CI 0.15–0.66), and female participants had lower odds than male participants (AOR 0.19; 95% CI 0.04–0.92). Conclusion: Stronger overall health beliefs were associated with higher self-efficacy, whereas higher knowledge was unexpectedly associated with lower self-efficacy. Because the design was cross-sectional and the HBM was analyzed as a total score, these findings do not establish causality or indicate which HBM components are most important. TB education may need to address patients' health beliefs, not only factual knowledge, and this approach should be tested in longitudinal and intervention studies.