Although online learning is rapidly expanding in the community college setting, there is little evidence regarding its effectiveness among community college students. In the current study, the authors used a statewide administrative data set to estimate the effects of taking one's first college-level math or English course online rather than face to face, in terms of both course retention and course performance. Several empirical strategies were used to minimize the effects of student self-selection, including multilevel propensity score. The findings indicate a robust negative impact of online course taking for both subjects. Furthermore, by comparing the results of two matching methods, the authors conclude that within-school matching on the basis of a multilevel model addresses concerns regarding selection issues more effectively than does traditional propensity score matching across schools. (Contains 1 figure, 3 tables, and 13 notes.)