Abstract: This work entails identifying areas of all coal pits in and around a part of Lower Benue trough where coal exploitation had taken place in the past, using random forest machine learning technique with out-of-bag (OOB) estimate of generalization error. Training sites were selected from the Landsat image covering the area of interest using QGIS image processing software. Two well-known exploited sites were selected as coal training sites while few other ones also identified were reserved for validation. Some other areas known to be different than coal sites were selected and classed as non-coal areas.