Background: Air pollutants are considered carcinogenic to humans. In some European countries, an association between industrial air pollution and childhood cancer has been established. This relationship has not been addressed in Latin America, despite the spatial variability of air pollutants that may limit the extrapolation of the results to other geographical areas. Objective: To conduct a spatial analysis of the relationship between childhood cancer and proximity to industrial sources of air pollution in a metropolitan area of Colombia. Methods: Incident cases of childhood cancers were obtained from the Population-based Cancer Registry of the Metropolitan Area of Bucaramanga during 2000-2015. Local and focused cluster tests were used for the detection of spatial clusters, and the Poisson multivariable model was used to evaluate the combined effects of spatial variables. Results: The Kulldorff's focused test found a significant spatial cluster (P < 0.001) around one industrial agglomerate and the multivariable model results suggests that the distance effect is modified by the directional effect of the wind. Conclusion: A spatial cluster of incident cases of childhood cancer occurred in the municipality of Bucaramanga, Colombia. Our finding supports the hypothesis that childhood cancer might be related with industrial air pollution exposure in a Latin American city.