Martha K. Huggins 
AT THE TURN OF THE SECOND MILLENNIUM, VISIBLE STATE-SPONSORED ATROCITIES in Brazil may seem to be only a pale reflection of the ethnic and political cleansing and other large-scale, often genocidal, violence in places such as Guatemala, East Timor, Rwanda, the Sudan, Bosnia, Kosovo, and Chechnya. The extent and scope of violence in these countries -- with its raw brutality, military context, and state and overt paramilitary state-sponsored murder and mistreatment of civilians -- make contemporary peacetime violence in Brazil, or even during its 21-year military dictatorship (1964 to 1985), appear to be much less noteworthy and significant. Yet, that assumption suggests the questionable moral thesis that smaller numbers or a lower percentage of victims out of the total Brazilian population reduce the significance of this and similar situations of state-permitted or encouraged violence. Whether defensible or not, the validity of such a proposition cannot be considered without examining the extent and nature of violence in Brazil, where interpersonal violence is prevalent and dramati c, although disguised because it is ethnically and class selective as well as geographically focused.
To indicate the extent of violence in Brazil, in Sao Paulo alone (the largest city in South America), between 1984 and 1996 there were 69,700 homicides -- over 10,000 more deaths than known U.S. casualties during the entire Vietnam War. Between 1979 and 1997, the homicide rate in Brazil increased from 11.5 murders per one hundred thousand population to 25.4. This 18-year period spans the six years before the military left power through the 12 years after formal democratic government had returned. Between 1979 and 1997, Brazil's population increased 65%, but its homicide rate went up 120% (Folha, 1999b: 3-1). In contrast to Brazil's 1997 homicide rate of 25.4, the corresponding U.S. homicide rate was 10.1 per 100,000 population (Internet: www.IABD.org), which then steeply declined in 1998 to 6.3 (Internet: www.ojp.usdoj.gov/bjs.homicide/hmrt.htm). By the 1990s, Brazil's homicide rate approached that of countries recently riven by militarized internal conflict. In the 1990s, Brazil had the fifth highest murder rate in the world, following Guatemala, El Salvador, Colombia, and Jamaica (Buvinic et al., 1999), with the latter also a country under nonmilitary but militarized social control.
Although Brazil's countrywide homicide rate  is already high by most international comparisons, the homicide rates in some of its districts and for some population sectors are equal to or even higher than Latin American countries still involved in internal guerrilla and military conflict. For example, Colombia -- formally democratic, yet in a militarized drug and guerrilla war with large regions of the country under military, paramilitary, or guerrilla control -- had a homicide rate in the early 1990s of 89.5 murders per 100,000 population (ibid.), clearly higher than Brazil's 1997 countrywide rate of 25.4 (NEV, 2000). Yet the overall homicide rate was lower than the 1998 rates for Sao Paulo City's Diadema district (140), where the auto industry has been downsizing amid slums and poverty, and the city's poor Embu district (97.32) (Folha, 1999b: 3-3). Colombia's rate was also lower than the cities of Recife (105) in Brazil's impoverished northeast, and Vitoria (103), in Espirito Santo state east of Rio (NE V, 2000). Indeed, Diadema's homicide rate is closer today to Guatemala's and El Salvador's national murder rates (150) in the late 1980s (Buvinic et al., 1999), when those countries were experiencing intense internal guerrilla and counterinsurgency conflict.
Brazil's pattern of urban violence is clearly most dramatic for selected regions and segments of the population. In 1997, the country's two largest cities, Rio de Janeiro and Sao Paulo, ranked third and fourth after …