We propose a tractable method for the estimation of a consideration set model in which consideration sets are endogenously determined through search. We show that the widely- used alternative-specific consideration model is a special case in which consumers put zero weight on expected utility when making their search decisions. To deal with the dimensionality problem that may arise from a large number of consideration sets, we propose a novel, accurate, and computationally fast Monte Carlo estimator for the choice probabilities. We use several existing datasets to identify the extent to which search played a role when buyers formed their consideration sets.