It is often difficult to obtain valid estimates of comparative treatment effectiveness and safety owing to differences across patient populations taking different medications in the real world. One approach for assessing comparability between treatment groups in effectiveness studies is to use negative control outcomes (NCOs). NCOs share similar sources of bias with the primary outcomes but have no plausible causal relationship to the treatment of interest. Observing differences in the risk of NCOs thus provides evidence for residual confounding between groups. This retrospective study assessed the comparability of postmenopausal women, treated with osteoporosis medications with various mechanisms of action such as denosumab (receptor activator of nuclear factor κB ligand [RANKL] inhibitor), zoledronic acid (bisphosphonate derivative), or oral bisphosphonates including alendronate. Administrative claims data were extracted from the US Centers for Medicare and Medicaid Services' Chronic Condition Warehouse database (May 2010–December 2016). Propensity scores were used to match denosumab patients 1:​1 to comparators. Four nonfracture NCOs and three early fracture NCOs (before substantial biologic effects of treatment would be expected) were assessed over 1-year and 3-month follow-up periods, respectively. According to comparability decision rules established a priori, patients initiating denosumab were comparable to those initiating zoledronic acid or alendronate, irrespective of prior osteoporosis treatment experience. Among new users, new switchers, and in the historical fracture subgroup, no meaningful differences were observed in the cumulative incidence of the seven NCOs comparing denosumab to zoledronic acid. This empirical examination can assist in the selection of appropriate comparator groups for future comparability research using real-world data.