Native advertising, which matches the look and feel of unpaid news and editorials, has exploded online. The Federal Trade Commission has long required advertising to be clearly and conspicuously labeled, and it recently reiterated that these requirements apply to native advertising. We explore whether respondents can distinguish native advertising and “regular” ads from unpaid content, using 16 native ads, 5 “regular” ads, and 8 examples of news/editorial content, drawn from multiple sources and platforms. Overall, only 37% of respondents thought that the tested examples of native advertising were paid content, compared to 81% for “regular” advertising, with variation by platform, advertiser, and labeling. Modest labeling changes materially increased the number of respondents that correctly recognized that native ads are paid content – but even these improved results fell well short of those for “regular” advertising. We also explored labeling preferences and self-reported concern about native advertising. Our findings indicate that native advertising involves a significant risk of deception which self-regulation has not addressed.