In the academic debate, Brazilian unionism has been characterized both as representative and as non-representative – with defenders of both positions grounding their arguments in the same basic facts. Taking issue with these interpretations, the article questions whether or not union membership is a suitable measure of representation. If, given Brazil’s corporative union structure, unions do not need to recruit members, to what extent can membership serve as an adequate measurement of union representation? Precisely what do union membership rates measure, and precisely what does it mean to "represent workers" in Brazil? The article first offers a conceptual discussion of the relationship between union membership and representation. The thesis is that in Brazil the former does not suffice as a measurement of the latter. But if the union membership rate is high, what is it measuring in any case? The second part of the text explores the empirical side of the question, through analysis of a 1994 survey conducted within four occupational categories. It becomes clear that membership rates are indicative of the degree to which the union makes its presence felt in the life of those it represents. But it also becomes clear that in and of itself this rate does not account for the complexity of the relations between the rank-and-file and union leaders. In short, the referent is blurry – but it is there. The text seeks to lend it greater conceptual and empirical clarity.