doing research on mortality and on migration to the Amazon.In 1964 the Brazilian Armed Forces seized power from the constitutionally elected President Joao Goulart. The end of civilian control marked the beginning of a series of aggressive economic policies designed to control inflation and to stimulate economic growth.' These measures contributed to the socalled Brazilian &dquo;economic miracle&dquo; in which the rate of increase in the real GDP rose to an average of about ten percent per year between 1969 and 1972. Among the policies adopted in the post-1964 period was the strategy of severely curtailing the power of labor unions as independent bargaining agents, and of consistently lagging wage increases behind rises in the cost of living. To keep labor costs low, the legal minimum wage was revised at regular intervals on the basis of elaborate mathematical formulations. While these readjustments were putatively intended to recover lost income eroded by inflation, the method was deliberately designed to provide less than full compensation (see DIESSE, 1974;. The consequent decline in the real minimum wage meant a deterioration in the standard of living for a large sector of the population. This study investigates the social consequences of the pattern of recent Brazilian economic growth by analyzing the relationship between the decline in real wages and the upsurge in infant mortality rates in two of the country's most important urban centers.' Sao Paulo, with a population of eight million in 1970, is the largest city in Brazil, and Belo Horizonte, with close to two million inhabitants, ranks third. The scope of the analysis is restricted to these two 'For a discussion of these policies and the political context within which they took place, see Baer and Kerstenetzky (1972( ), Schmitter (1973, Skidmore (1973), and Singer (1972). 2 Other studies have suggested that the rise in infant deaths was caused by a decline in a real minimum wage (see Suplicy, 1974; Singer, 1974; Hoffman, 1971; Yunes andRonchezel, 1974 andBacha, 1976), but this association has not been systematically analyzed so as to account for other variables that could interfere with the relationship.