Objectives To evaluate trends in premature cancer mortality in Colombia by educational level in three periods: 1998–2002 with low healthcare insurance coverage, 2003–2007 with rapidly increasing coverage and finally 2008–2012 with almost universal coverage (2008–2012).

Setting Colombian population-based, national secondary mortality data.

Participants We included all (n=188 091) cancer deaths occurring in the age group 20–64 years between 1998 and 2012, excluding only cases with low levels of quality of registration (n=2902, 1.5%).

Primary and secondary outcome measures In this descriptive study, we linked mortality data of ages 20–64 years to census data to obtain age-standardised cancer mortality rates by educational level. Using Poisson regression, we modelled premature mortality by educational level estimating rate ratios (RR), relative index of inequality (RII) and the Slope Index of Inequality (SII).