There are about three million cases in a national voter fie in which 2012 vote records share a common first name, last name, and date of birth. We develop a probabilistic birthdate model to estimate how many of these cases represent the same person voting twice. If registration records are never erroneously marked as being used to vote, we estimate about 0.02% of the votes cast in 2012 were double votes. An audit of poll books, however, suggests that such measurement error could explain many of these apparent double votes. Using data returned to Iowa by the Interstate Crosscheck Program, we quantify the tradeoff between voter accessibility and electoral integrity when purging a likely duplicate registration from another state. We find that one of Crosscheck's proposed purging strategies would eliminate about 300 registrations used to cast a seemingly legitimate vote for every one registration used to cast a double vote.