This study addresses the impact of Mexican illegal immigration on the U.S. labor market. It constitutes a first step towards developing rigorous structural econometric models that empirically analyze undocumented labor force dynamics. Structural estimation of the labor supply and the participation decision of illegal Mexican immigration requires the solution of intricate theoretical problems that have not been addressed in previous literature. The analysis developed here identifies those problems and proposes innovative solutions. In particular, undocumented participation in the U.S. labor market is studied in the context of life cycle theory and stochastic behavior. The empirical part of the analysis reviews the problems of sample selection and missing observations that characterize the available data on Mexican migration. The proposed empirical specification is evaluated employing limited dependent variables procedures, where a Tobit simultaneous equation model is solved using maximum likelihood methods. According to the empirical results, Mexican undocumented immigration may be viewed as a transitory phenomenon. Individuals switch back and forth between Mexico and the U.S. reacting not only to income differentials, but also to social, family and economic attachments in their home-communities. Mexican workers seem to have little incentives to invest in human capital specific to the U.S., such as the ability to speak English. This behavior may be result of the partial transferability of Mexican skills, i.e. formal education, to the secondary market in the United States. Finally, contrary to conventional wisdom, the empirical evidence suggests that exogenous increases in U.S. wages, i.e. a non-expected hike in the legal minimum wage, may actually discourage Mexican undocumented participation in the U.S. labor market.