Why do some democracies adopt effective development policies while others remain mired in stagnation or suffer cyclic crises? Previous studies have emphasized poverty or institutional weakness, but weak accountability and development failure are not limited to low-income countries or to any specific institutional choices. A better explanation, Mona Lyne argues, is provided by her theory of "the voter's dilemma": where structural conditions render quid pro quo, or clientelistic, politics viable on a national scale, voters have insufficient incentive to support politicians promising national public goods policies. Under these conditions, Lyne argues, electoral accountability falls prey to the same n-person prisoner's dilemma that plagues any other large-scale decentralized attempt to procure collective goods. The theory is tested through an examination of four prominent cases. A comparison of postwar Brazil and pre-Chavez Venezuela shows that clientelism debilitated both countries' postwar development programs, despite Venezuela's historically strong institutions and abundant oil revenues. Two comparisons-one between contemporary Brazil and pre-Chavez Venezuela, and another between postwar and contemporary Brazil-highlight factors that reduce the risks of rejecting clientelism as providing the best account of contemporary Brazil's success. Finally, a comparison of pre-Chavez and contemporary Venezuela explains the continuity in flawed institutional and policy choices as a result of continuity in clientelistic politics driven by the voter's dilemma.