To describe a measure of drug compliance on each patient at multiple time points, and illustrate an approach for quantifying the impact of covariates on these longitudinal compliance measurements. Nine thousand hypertensive patients from MEDSTAT's Marketscan database were identified who filled an antihypertensive drug 5 times in 2006. At each fill, days since expiration of previous fill's days supply (gap) was measured, a smaller gap suggesting better compliance. Also measured were covariates such as demographics, co-medication, and co-pay. A random effect model was used to describe the dependence of the five gap compliance measurements on each patient, with a random intercept reflecting the impact of both measured and unmeasured covariates. Patient specific slopes were used to reflect relationships between gap and covariates. The average gap was 5 days and the fraction of total gap variance due to heterogeneity across patients was 27%. The gap within patients increased on average over time (0.5 day increase in gap size per month) but this varied considerably between patients (SD=0.8). The gap was larger for diuretic than for non-diuretic drug users, decreased slightly with age, and was not associated with co-pay. Unlike single measures of compliance, longitudinal measures provide opportunity to describe patterns over time within patients, and permit impact of changing covariates within patients to be estimated.