How does the state respond to members of the public seeking to mobilize its coercive power? Focusing on welfare fraud control units in the United States, we examine how race/ethnicity and written English proficiency affect access to systems for reporting welfare fraud suspicions. Using a correspondence audit, we assess fraud control authorities’ likelihood of taking up reports from Latinas and Whites with higher and lower English proficiency. We find that fraud units are less likely to take up lower-proficiency Whites’ reports, but that lower proficiency’s uptake-dampening effect does not hold for Latinas. To explain the mechanisms underlying our experimental results, we draw on interviews with fraud investigators. The interview evidence reveals the determinations of investigative promise underlying these uptake disparities. For White reporters, English errors cue gatekeepers’ preexisting skepticism about public reporters’ reliability, decreasing enthusiasm for investing resources in these reports. Reports from lower-English proficiency Latinas offer special viability appeal, however, offsetting the negative influence on uptake probability that errors demonstrate for White reporters. Our results shed new light on contemporary racial/ethnic dynamics in the US welfare system, and advance social scientific understanding of why some people—and not others—become consequential contributors to the exercise of state power.