OBJECTIVES Compare respiratory-related costs between cohorts of compliant and non-compliant patients with chronic obstructive pulmonary disease (COPD) initiated on maintenance therapy (tiotropium or fluticasone/salmeterol). METHODS This retrospective cohort analysis used medical and pharmacy claims from a large national U.S. health plan. Patients were commercial health plan enrollees ≥40 years old with ≥1 tiotropium or fluticasone/salmeterol fill between December 1, 2004 and December 31, 2005, COPD, and ≥12 (up to 18) months continuous enrollment. Patients had no fills of tiotropium, fluticasone/salmeterol, salmeterol, or formoterol during a six-month pre-index period. Cohorts were "compliant" (medication possession ratio [MPR]≥80%) or non-compliant" (MPR<80%). Respiratory-related inpatient costs, medical costs, and total (medical + outpatient pharmacy) costs were modeled with generalized linear model (GLM) with log link and gamma distribution. Costs were adjusted to 2006 dollars. Covariates included cohort, demographics, and COPD severity, index pharmacy claim from pulmonologist (intensity of services), and comorbidities. RESULTS The sample included 558 (12.3%) compliant and 3979 (87.7%) non-compliant subjects (total N=4537). Compliance compared with non-compliance was associated with: higher total respiratory-related costs (cost ratio=1.47, 95% confidence interval [CI]=1.13-1.90, p<0.01); lower respiratory-related medical costs (cost ratio=0.64, CI= 0.44-0.92, p<0.05); and lower respiratory-related inpatient costs (cost ratio=0.49, CI=0.32-0.75, p<0.01). COPD severity, baseline oxygen use, and pulmonologist index claim were significantly associated with higher respiratory-related total costs and medical costs. When treatment (tiotropium vs. fluticasone/salmeterol) was added as a covariate, compliance remained significant and treatment was not. CONCLUSIONS Compliance was associated with higher total respiratory-related costs, because sustained therapy was correlated with higher respiratory-related outpatient pharmacy costs, and likely also with higher respiratory-related medical services consumption. However, compliance was also associated with lower respiratory-related medical and inpatient costs. Costs were not linked directly with clinical outcomes, but outcomes may be indirectly inferred from medical and inpatient costs. Further research should examine compliance with respect to outcomes and costs.