Masonry is not only one of the significant ancient construction materials but also has wide usability as a modern material owing to its modest construction technique and economy compared to other prevalent building materials. Though significant research has been conducted to address the mechanical traits of masonry still there is lack of robust methods in the literature to predict its strength. Due to the nonlinear relationship between the constituents of masonry. In this investigation, models using machine learning are developed for determining the masonry prisms’ strength due to compression. The various models including the M5P model tree, multilayer perceptron and random forest regression have been augmented to obtain the best-fit prototype using a 10-fold cross-validation technique. The database composed of 132 datasets was generated. 80% of the complete dataset was employed to train the model and the residual 20% for testing the models. The results revealed that Random Forest performed preferably to the other two models in regards to its capacity for prediction. To wholly comprehend the behaviour of different inputs on the predicted strength due to the compression of prisms, a sensitivity analysis is done. Findings reveal that the primary variables influencing strength due to compression are the mortar strength (fm) and the ratio of mortar in horizontal joints to mortar in both joints of the masonry prism (VRmH).