Virtually all legislative theory rests upon the assumption that politicians are driven by the desire to win repeated reelection. Indeed, because it is so often taken at face value, John Carey recently noted that the reelection assumption “has reached near axiomatic status” (1994, 127) among political scientists. It is important to understand that this assumption implies not only that legislators direct their energies toward ensuring repeated reelection, but that they usually succeed in their efforts. All else equal, we expect little legislative turnover in systems where the reelection assumption holds.

At first glance, Brazil appears to be a case that confirms this assumption's validity. As in the United States, Brazilian incumbents do not require national party leaders' approval to run for reelection. Moreover, Brazil's electoral laws actually encourage incumbency. Incumbents do not have to battle to win renomination, because a “birthright candidate” (candidato nato) law automatically places their names on the next election's ballot (until 2002). Given this institutional backdrop as well as the idea's intuitive plausibility, several scholars have employed the reelection assumption to explain important aspects of Brazilian – and comparative – politics (e.g., Ames 1987, 1995a; Geddes 1994).

Yet upon closer examination Brazil turns out to be a particularly perplexing case. Although its electoral laws encourage incumbency, in contrast to the United States (where turnover in the House is less than 10 percent with each election) turnover in the Brazilian Chamber of Deputies has consistently exceeded 50 percent.