An increasingly fragmented media environment poses a challenge to campaigns and political organizations trying to persuade young voters, as young people increasingly eschew television for online video streaming and represent a growing cell-phone-only population, which is more costly to reach. As a result, each cycle campaigns are spending more on online advertising. However, the effectiveness of this outreach is an open question with disparate findings in the literature (Bond et al. 2012, Broockman and Green 2012). To test the effectiveness of this outreach for voter mobilization, we partnered with a national organization, Rock the Vote, to conduct a seven-state, 731,568-person GOTV experiment using advertising on Facebook during the 2012 presidential election, which involved nearly four million advertising impressions, and a 14-state, 93,053-person replication during the 2013 general election. Across both experiments, we find no evidence that voter turnout in the treatment group was greater than the control group. While exploratory analysis from the 2012 experiment suggested the ads were effective among those who had a demonstrated a prior history of clicking on online advertising, the follow-up 2013 experiment failed to confirm this finding. These results suggest that while online advertising can reach young voters, it may not be a panacea for encouraging them to vote. This research also underscores the importance of replications of even large-scale experimental findings.