Quota mechanisms are commonly used to elicit private information when agents face multiple decisions and monetary transfers are infeasible. As the number of decisions grows large, quotas asymptotically implement the same set of social choice functions as do separate mechanisms with transfers. We analyze the robustness of quota mechanisms. To set the correct quota, the designer must have precise knowledge of the environment. We show that, without transfers, only trivial social choice rules can be implemented in a prior-independent way. We obtain a tight bound on the decision error that results when the quota does not match the true type distribution. Finally, we show that in a multi-agent setting, quotas are robust to agents’ beliefs about each other. Crucially, quotas make the distribution of reports common knowledge.