the objective was to identify hospital admission associated factors in brazil, analyzing if that utilization is equitable and identifying which characteristics are associated with heavy users. we analyzed data from the 1998 national household survey, using logistic regression and multinomial logistic regression, with normalized weights and statistical techniques to correct for design effect. we used andersen's behavioral model as the theoretical model for this analysis. in model adjusted for health needs and enabling factors, people with smaller income had more chance of being admitted to hospital; and in model adjusted only for health needs people with larger income had more chance. all need variables were less related to use for people with two hospital admissions, in comparison with those with more than two admissions; and there was no association between social variables and occurrence of two admissions, but this association occurred for three or more admissions. the reduction of social inequalities in the aspects that "enable" people to use hospital admissions would reduce inequalities in this use. a health system that offers a "regular health service", besides low or any payment in the moment of use, would have a positive impact in the equity of hospital admission utilization in brazil.