Although medication adherence can be measured many ways, researchers often view electronic dose monitoring devices like the Medication Event Monitoring Systems (MEMS) as more valid than patient self-reports. MEMS are popular but have potential problems. Based on the literature and MEMS data analyses, the authors suggest the following approaches: (a) a 1- to 2-month run-in should be used to reduce MEMS reactivity, (b) MEMS should be used with other measures of adherence, (c) adherence should be measured continuously, or a cutoff should be defined based on pharmacological properties of the medication and the consequences of nonadherence, (d) MEMS data usually should be aggregated weekly or monthly and evaluated using multilevel modeling, (e) MEMS-based screening for nonadherence may miss some patients in need of intervention, and (f) researchers should use strategies like training patients to use MEMS and purchasing extra MEMS caps to improve the completeness and accuracy of MEMS data.