Ibrutinib is used to treat B-cell malignancies. Bleeding occurred in up to 50% of ibrutinib-treated clinical trial participants, but major bleeding was uncommon. However, trials excluded patients with certain risk factors and may underestimate major bleeding frequency in clinical practice. Furthermore, many ibrutinib-treated patients concomitantly use oral anticoagulants (OACs) to prevent thromboembolic events caused by comorbid conditions and ibrutinib-related atrial fibrillation. Thus, clinicians are confronted with the decision of concomitantly treating patients with ibrutinib and OACs without sufficient knowledge of the bleeding risk. We conducted two cohort studies in a commercially-insured population to estimate hazard ratios (HRs), using inverse probability of treatment weighting-adjusted Cox proportional hazards regression, for bleeding with ibrutinib vs. bendamustine-rituximab (BR). We also evaluated the performance of SuperLearner-high-dimensional propensity score (SL-hdPS), a novel approach that reduced propensity score overfitting and improved bias reduction vs. hdPS alone in simulation studies. In Aim 1, we assessed rates of major (resulting in hospitalization) and clinically-relevant (resulting in emergency department presentation or hospitalization) bleeding with ibrutinib versus BR in individuals not concomitantly treated with OACs. The major and clinically-relevant bleeding HRs for ibrutinib (vs. BR) were 1.61 (95% confidence interval: 0.67-3.84) and 2.66 (1.29-5.49), respectively. In Aim 2, we assessed rates of major, clinically-relevant, and provider-diagnosed (in any healthcare setting; specified post-hoc) bleeding with ibrutinib vs. BR in individuals concomitantly treated with OACs. Major and clinically-relevant bleeding HRs could not be estimated due to low event counts. The provider-diagnosed bleeding HR for ibrutinib + OAC (vs. BR + OAC) was 2.70 (1.25-5.82). In Aim 3, we evaluated the performance of SL-hdPS compared to hdPS alone using the Aim 1 and 2 cohorts. SL-hdPS resulted in better covariate balance than hdPS alone in the Aim 1 cohort but resulted in overfitting and suboptimal covariate balance in the Aim 2 cohort. This work fills an important knowledge gap in ibrutinib-related bleeding which may inform ibrutinib’s benefit-harm balance. It also provides support for the use of SL to help optimize covariate balance when using hdPS in certain small sample studies.