In Brazil, where the use of pesticide grows rapidly, studies that evaluate the impact of pesticide exposure on cancer incidence and mortality are very scarce. In this study, we evaluated the degree of correlation between pesticide sales in 1985 in eleven Brazilian states and cancer mortality rates during 1996-1998. Information of all cancer deaths occurred in men 30-69 years old from 1996 to 1998 were collected from National Mortality System. Single and multiple linear regression coefficients were obtained to assess the relationship between per capita sales of pesticides in 1985, specific-site cancer mortality rates (prostate, soft tissue, larynx, leukemia, lip, esophagus, lung, pancreas, bladder, liver, testis, stomach, brain, non-Hodgkin's lymphoma, and multiple myeloma) during 1996-1998, and several covariates. In addition, states were stratified into three groups according to tertiles of pesticides sales and cancer mortality rate ratios (MRR) were then calculated using first tertile as reference. Finally, a factor analysis was performed to reveal unapparent relationships between pesticide use and cancer mortality. Pesticide sales showed statistically significant correlation with the mortality rates for the cancers of prostate (r=0.69; p=0.019), soft tissue (r=0.71; p=0.015), leukemia (r=0.68; p=0.021), lip (r=0.73; p=0.010), esophagus (r=0.61; p=0.046), and pancreas (r=0.63; p=0.040). Moderate to weak correlations were observed for the cancers of larynx, lung, testis, bladder, liver, stomach, brain, and NHL and multiple myeloma. In addition, correlation between pesticide sales and specific-site cancer mortality rates was reinforced by multiple regression analysis. For all specific-sites, cancer mortality rates were significantly higher in the states of moderate (2nd tertile) and high (3rd tertile) pesticide sales, with MRR ranging from 1.11 to 5.61. Exploring hidden relationships between pesticide sales and cancer mortality in Brazil, through a factor analysis, revealed that affluence; public policies and lifestyle behaviors may explain almost 70% of the variance of the studied association. The results suggest that population exposure to pesticides in the 1980s in some Brazilian States may have been associated with selected cancer sites observed a decade later.