We examine the physical and mental health effects of providing care to an elderly mother on the adult child caregiver. We address the endogeneity of the selection in and out of caregiving using an instrumental variable approach, using the death of the care recipient and sibling characteristics. We also carefully control for baseline health and work status of the adult child. We explore flexible specifications, such as Arellano–Bond estimation techniques. Continued caregiving over time increases depressive symptoms and decreases self-rated health for married women and married men. In addition, the increase in depressive symptoms is persistent for married women. While depressive symptoms for single men and women are not affected by continued caregiving, there is evidence of increased incidence of heart conditions for single men, and that these effects are persistent. Robustness checks indicate that these health changes can be directly attributable to caregiving behavior, and not due to a direct effect of the death of the mother. The initial onset of caregiving has modest immediate negative effects on depressive symptoms for married women and no immediate effects on physical health. Negative physical health effects emerge 2 years later, however, suggesting that there are delayed effects on health that would be missed with a short recall period. Initial caregiving does not affect health of married men. Published in 2009 by John Wiley & Sons, Ltd.