OBJECTIVES: The relationship between inequities in socioeconomic factors (i.e. poverty) and adverse health outcomes has being studied before, but most studies were aggregated by country. A further disaggregation may contribute to causality and may help explain the within country differences and lack of improvements of some health indexes. In order to assess inequalities in Colombia this study evaluate by geographical region (state) of the infant mortality rate in Colombia in 1993 and 2005, the two years where the census took place. METHODS: This ecological study measures regional inequalities in the infant mortality rate (IMR), using these indexes: Rate difference (RD), incidence rate ratio (IRR) and population attributable risk (PAR). Poverty was measured as the index of unsatisfied needs (Necesidades Básicas Insatisfechas, NBI). A linear regression was made for the association between IMR and NBI (poverty), and beta-coefficients with 95% confidence intervals (95%CI) were estimated.  RESULTS: In 1993, the absolute difference between regions (states) with more and less poverty (Chocó: 80.4%; Bogotá, D.C.:17.3%) was 63.1%. In 2005, this difference was 70.4%.  For these years, the IRR of IMR between those (Choco and Bogota) states was 2.9 y 4.2 and RD 56.4 y 59.2, respectively. The R-square between poverty and IMR was in 2005 63.8%, versus a R-square of 34.2% for 1993. The beta-coefficient of linear regression was 0.339 for 1993 (IC 95%, 0.167-0.552). For each perceptual point that decreases poverty in the states, a reduction of TMI by 0.339 for each 1,000 live births. In 2005, this reduction was 0.646 for each 1,000 live births (IC 95%, 0.468-0.823). CONCLUSIONS: Although Colombia has a trend of TMI reduction, health inequalities persist between states in the country. These inequalities increased between 1993 and 2005.