This study examined associations between select hypothesized adherence predictors in claims databases and long-term inhaled corticosteroid adherence. Medication adherence was measured by using medication possession indexes and a refill regularity measure in 1595 older adults using inhaled corticosteroids for 2 years. Medication possession indices were highly correlated with each other but not with the refill regularity measure. Additionally, the hypothesized predictors explained a larger percentage of the variance in the medication possession indices compared to the refill regularity measure. Although long-term retrospective medication utilization poses many measurability issues, the use of multiple indices gives a more complete picture of usage behavior.