Adherence to pulmonary rehabilitation (PR) is low. Previous studies have focused on clinical predictors of PR completion. We aimed to identify social determinants of adherence to PR. A cross-sectional analysis of a database of COPD patients (N = 455) in an outpatient PR program was performed. Adherence, a ratio of attended-to-prescribed sessions, was coded as low (<35%), moderate (35–85%), and high (>85%). Individual-level measures included age, sex, race, BMI, smoking status, pack-years, baseline 6-minute walk distance (6MWD: <150, 150–249, ≥250), co-morbidities, depression, and prescribed PR sessions (≤20, 21–30, >30). Fifteen area-level measures aggregated to Census tracts were obtained from the U.S. Census after geocoding patients' addresses. Using exploratory factor analysis, a neighborhood socioeconomic disadvantage index was constructed, which included variables with factor loading >0.5: poverty, public assistance, households without vehicles, cost burden, unemployment, and minority population. Multivariate regression models were adjusted for clustering on Census tracts. Twenty-six percent of patients had low adherence, 23% were moderately adherent, 51% were highly adherent. In the best fitted full model, each decile increase in neighborhood socioeconomic disadvantage increased the risk of moderate vs high adherence by 14% (p < 0.01). Smoking tripled the relative risk of low adherence (p < 0.01), while each increase in 6MWD category decreased that risk by 72% (p < 0.01) and 84% (p < 0.001), respectively. These findings show that, relative to high adherence, low adherence is associated with limited functional capacity and current smoking, while moderate adherence is associated with socioeconomic disadvantage. The distinction highlights different pathways to suboptimal adherence and calls for tailored intervention approaches.