To study the variations in infant mortality rate (IMR) across Colombia's 33 administrative departments over the period 2003-2009, examine persistency of variations across departments over time, and relate those variations to the impact of socio-economic conditions and availability of care on IMR.Using vital statistics and related socio-economic data we establish three types of analysis according to: (a) the variation of the departmental IMR (2003-2009), (b) the association between the departmental IMR and its key determinants over time, and (c) the lines of causality and relative impact of different factors, by using structural equations.The 4.7 fold ratio between the highest and lowest departmental IMR (2009) may be underestimated considering underreporting, especially in low-income departments. There is a negative association between the departmental IMR with time and a set of highly correlated variables, such as the mother education, income per capita, health insurance level and access to services.The effect of better insurance, availability of private beds, and having doctors attending mothers, eclipse the impact of better socioeconomic conditions. The range of services does not appear to be influenced by a rational policy; resources are not allocated according to the need, but with the general development. Private beds are made available where there is better health insurance.