Scholars often use roll-call votes to study legislative behaviour. However, many legislatures only conclude a minority of decisions by roll call. Thus, if these votes are not a random sample of the universe of votes cast, scholars may be drawing misleading inferences. In fact, theories over why roll-call votes are requested would predict selection bias based on exactly the characteristics of legislative voting that scholars have most heavily studied. This article demonstrates the character and severity of this sampling problem empirically by examining European Parliament vote data for a whole year. Given that many legislatures decided only a fraction of their legislation by roll call, these findings have potentially important implications for the general study of legislative behaviour.