Abstract: In this paper, an intelligent identification method was proposed to ascertain overlying strata failure types and fracture development in the overlying strata of a goaf, and identify three zones (sagging zone, fractured zone and caving zone) quickly and accurately. This method is expected to be a basis in later development of goaf treatment plans. A coal mine goaf in Shandong Province was introduced for case study, where the Bayesian Online Changepoint Detection (BOCD) algorithm was employed to analyze the changes and response characteristics of the drilling fluid loss and drilling rate over the drilling process at the boundaries of three zones. With exploration specification empirical formulas for calculating the heights of three zones as constraints, the candidate changepoints were detected and optimized from the drilling fluid loss and drilling rate data. Then, the boundary depths of three zones of the goaf were determined. The intelligent identification results are in good coincidence with the actual values. Specifically speaking, the depth errors of the lower boundaries of sagging zone, fractured zone and caving zone are +0.67 m, +0.31 m, and +0.52 m, and the height errors of sagging zone, fractured zone and caving zone are +0.14% and −0.63% and +2.49%. The intelligent identification method for overlying strata three zones of a goaf based on drilling data is demonstrated to be acceptable in accuracy and available for use. The proposed method presents a good combination of drilling data with empirical formulas, which permits boundary division of three zones during drilling, full play of data timeliness and elimination of impacts arising from technicians’ subjective judgment on the identification results. Superior to conventional techniques referring multiple methods to identify the boundaries of three zones, the intelligent identification method significantly enhances the timeliness and accuracy of three zones division.