OBJECTIVES: To compare several adherence metrics across different therapeutic conditions, which might complicate these associations. METHODS: Plan participants dispensed lipid lowering, antihypertensive, or selective serotonin reuptake inhibitor between January 1, and June 30, 2004, had their same class prescriptions extracted for six months before and one year after their first prescription from a large deidentified US prescription database. Lipid-lowering (n=102,067), antihypertensive (n=139,333), and SSRIs (n=99,439) plan participants without a same-class prescription in the pre-period had: number of prescriptions -- prescription count adjusting 3-1 for mail-retail; persistent period -- first minus last fill date; length of exposure -- persistent period plus days supply on last prescription; total days -- sum of all days supply; unique days -- sum of days with exposure to drug; medication possession ratio -- sum of days supply divided by length of exposure (variable) or study period (fixed); proportion of unique days -- sum of days with exposure divided by length of exposure (variable) or study period (fixed); and continuous days -- days spanned before a break in therapy calculated over the post-period. Results for these adherence metrics were divided into equal parts for comparisons. RESULTS: The measures were grouped into: 1) MPR variable and PUD variable, 2) unique days 3) MPR fixed, PUD fixed, total days, length of exposure, and number of prescriptions; 4) persistent period; 5) continuous days. Differences among the groups started within the first 30 days, with group 1 having values less than 1 percent and the continuous group having more than 20%. Groups 1 and 2 remained lower than the other groups ending at about 50% while the others ended at about 70%. CONCLUSIONS: Differences in how adherence is calculated needs to be considered when estimating benefits. Other metrics that account for dose changes, concomitancy, and other complex regimens may need further exploration.