The Zarit Burden Interview (ZBI) is believed to be the most commonly used measure of caregiver burden. Originally developed more than 20 years ago for use with informal caregivers of community dwelling persons with Alzheimer disease, it has subsequently been administered to a diverse range of patient populations, formal or paid caregivers, and translated into numerous languages. Given that the ZBI is now used more broadly than it was initially intended and first validated, the current study applies the reliability generalization meta-analytic procedure to examine the psychometric properties of responses to the ZBI across populations. Multiple regression with categorical variables was performed to identify factors associated with error variance in ZBI reliability estimates (N=138 data points). Number of items, residence of the care recipient (community) and the Hebrew version each contributed significantly to prediction of internal consistency. These differences, however, were found to be relatively small and within accepted parameters. Generally, responses to the ZBI appear reliable across populations of caregivers and patients. Only versions of the ZBI with more or less than 22-items (nonstandard formats) reflect both statistical and meaningful differences in reliability. Where feasible, it is recommended that the 22-item version of the ZBI be used in future research and clinical practice.