Three methods for constructing standard errors of welfare estimates have been employed in the recreation demand literature: a Taylor's series approximation, the bootstrap, and a method proposed by Krinsky and Robb. This paper presents the results of a simulation experiment designed to examine the accuracy of these methods.