The predictions of different classic migration theories are tested by using incentivized laboratory experiments to investigate how potential migrants decide between working in different destinations. First, the authors test theories of income maximization, migrant skill-selection, and multi-destination choice as they vary migration costs, liquidity constraints, risk, social benefits, and incomplete information. The standard income maximization model of migration with selection on observed and unobserved skills leads to a much higher migration rate and more negative skill-selection than is obtained when migration decisions take place under more realistic assumptions. Second, these lab experiments are used to investigate whether the independence of irrelevant alternatives assumption holds. The results show that it holds for most people when decisions just involve wages, costs, and liquidity constraints. However, once the risk of unemployment and incomplete information is added, independence of irrelevant alternatives no longer holds for about 20 percent of the sample.