In this article I review the use of randomized field trials to evaluate school voucher interventions. I argue that although estimates of the effect of the voucher offer on achievement are unbiased in these trials, more specific interpretations such as the effect of attending private school may be difficult to obtain. I discuss several evaluation parameters in the context of voucher experiments, with particular attention to the theoretical assumptions needed to identify private school impacts. I argue that if there are direct effects of the voucher offer on student achievement, and compliance with voucher assignment is imperfect, instrumental variable estimates are likely to overstate the extent to which achievement gains may be attributable to private schooling. (Contains 3 tables, 2 figures and 12 footnotes.)