Two-year colleges are an important part of the higher education system in the United States but there are concerns as to how attendance at these institutions affects educational attainment and labor market outcomes. This paper uses data from a nationally representative survey to examine the impact of students beginning their college career at a two-year college instead of a four-year college. Treatment effects are estimated using both standard regression techniques as well as propensity score matching. As these estimates may be contaminated because of selection on unobservable characteristics this paper will also employ a number of sensitivity analyses to consider the potential bias. The results show large negative impacts on both educational attainment and labor market outcomes for men and women who begin at a two-year college, even for those students who expect to complete a bachelor's degree. The evidence from the sensitivity analyses suggest that to eliminate these large effects there would need to be substantial, and arguably implausible, selection on unobservable characteristics. (Contains 11 tables and 2 figures.)