ABSTRACT Objective: To analyze the spatiotemporal pattern and factors related to infant mortality in Northeastern Brazil from 2008 to 2018. Method: Ecological study developed with infant deaths that occurred in the Northeast and reported in the Mortality Information System. Non-spatial and spatial regression models were used to identify indicators related to infant mortality. Results: The mortality rate showed a decreasing trend of 2.1% per year (95% CI: -2.7 - -1.6; p<0.001), with higher Bayesian coefficients concentrated in in municipalities in the interior of Piauí. The variables related to infant mortality were: Gini Index (β = 6.56; p=0.01), Municipal Human Development Index (β = -22.21; p 0.001), dependency ratio (β = 0.16; p <0.001), percentage of people in households without electricity (β = -0.12; p<0.001) and percentage of women aged 10 to 17 who had children (β=0.19; p=0.01). Conclusion: There was a decrease in infant mortality during the studied period and high Bayesian rates in the interior of Piauí.