Aircraft structures are subjected to several types of manoeuvres during flight and experience variety of loads. Flight loads simulation is an indispensable and elaborate process during design and development phase, which is verified by subjecting aircraft and its components to structural integrity tests on ground. In developmental flight-testing, loads were assessed through strain monitoring and validated with flight loads simulation results. With the advancement in data analysis techniques, data driven models are developed to predict flight strain thereby manoeuvre loads, using flight parameter inputs gathered during flight. In this work, three of the popular Machine Learning (ML) algorithms such as Lasso Regression (LR), Neural Networks (NN), and Random Forest (RF) were applied to predict the strain on highly stressed components. For the pull up manoeuvre with gentle roll flight regime, ML algorithms were applied to estimate the strain on the simulation data set obtained from full aircraft flight loads generation. Likewise, the strain predicted from real flight test data is processed using aforementioned ML methods for alike flight regime. Using the simulation and real flight strains as metrics, the manoeuvre loads on real flight is predicted. It’s observed that the predicted flight loads, matches within 10% error margin, in particular on high vertical acceleration flight regimes. The performance of ML methods pertinent to the data filtering, training data size and feature importance is discussed in detail. Neural network and Random Forest algorithms exhibit better performance metrics on wing root components.