This study aims to validate the accuracy of the Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS) dataset at representing climate variability over various time scales. The study is carried out in Antioquia, in northwestern Colombia, and uses statistically independent information provided by 75 rain gauges located at different sites from 1981 to 2018. In this study, statistical metrics are used to analyze the error structure of the CHIRPS data. Resilient methods for validation are included based on Gauss-Markov diagnostic problems and influence points, leverage, and outliers in the form of the trimmed least squares boot method and principal component analysis. To represent the performance of CHIRPS more accurately at estimating precipitation at different time scales, a comparison is made using 12 specific rain gauge located in different subregions of Antioquia, which had well-differentiated climatic characteristics. The results show that the accuracy of CHIRPS estimates is conditioned by geographic and climatic characteristics of the region in which precipitation events occur. CHIRPS works well in most weather conditions in Antioquia, even during the most intense periods of ENSO. However, it does better in the Andean subregions and underperforms in warmer regions. This study finds that CHIRPS is a rainfall data source that can be considered for analyzing seasonal and interannual variability and spatial precipitation patterns. It provides good spatio-temporal coverage and is particularly suitable for areas with few rainfall gauges. However, its accuracy is minimal for daily analysis.