<p>One of the primary challenges in open pit mining is blast-induced ground vibration causing structural damages that require regular monitoring. The induced ground vibration levels may be restricted by recommending the maximum explosive charge per delay (Qmax) across a range of distances (D) to monitoring points. However, there is always a probability that the recommended blast design may exceed the desired PPV limits. In this study, a probabilistic approach using logistic regression was undertaken to classify dichotomous PPV outputs into two categories (&lt; 5 mm/s and &lt; 10 mm/s) using Q_max and D as input parameters. The model was trained and validated for both the cases using blast vibration monitoring data from chromite and iron ore mining regions, respectively. Model training and validation exhibited accuracy and precision of 81%, 90% and 91%, 92%, respectively (for &lt; 5 mm/s) and 77%, 90% and 90% and 92%, respectively (for &lt; 10 mm/s). Lastly, the Q_max in the chromite mining region was also optimized for threshold PPV limits as per the prescribed legislative guidelines. Thus, a framework for estimating the safe limits of ground vibrations due to open pit bench blasting operations was achieved through a probabilistic approach to overcome the limitations of deterministic prediction models.</p>