a cross-sectional study was conducted with 65 finishing pig herds from the states of rio grande do sul and santa catarina, brazil, in order to identify factors associated with seroprevalence for salmonella. pig farms were visited one week prior to slaughtering of animals when personnel were asked to answer a questionnaire. feed was also sampled for attempts of salmonella isolation, water for colimetrics analisis and around 40 pigs were bled. feed samples were subjected to salmonella isolation in selective media and sera were tested in an elisa, plates were coated with antigens of serovar typhimurium. herds were classified in one of three categories according to the prevalence of elisa positive sera, being low (less then 40% of positive sera), medium (between 40 and 70% positive sera) or high (more then 70% positive sera). seroprevalence was used as the explanatory variable and results obtained from the attempts to isolate salmonella from feed, water colimetrics results and the questionnaire answers were used as explanatory variables. initially, attempts of association between explanatory and explained variables were performed using the chi-square test. when associated (p￡0.1), the two variables underwent multiple correspondence analysis. factors associated with herds having high seroprevalence were: in finishing herds, pelleted feed, swine manure disposal less than 100m from surface water, feeder not provided with water drinker, swine from several herds transported together to slaughterhouse; in the farrow-to-finish herds, feed ingridients exposure to other animals, no active rodent control, dry feed, absence of fence, whitewashing of facilities after cleaning and disinfecting and permission for other people entrance to the herd. among the 65 pig herds visited, 98.5% were elisa-positive, with seroprevalence of 57.6% (confidence interval 56 to 60%).