The behaviour of masonry walls subjected to lateral loads is inherently complex, due to the anisotropic properties of masonry and nonlinear interactions between the masonry concrete block, mortar, grouted cells, ungrouted cells, and reinforcing steel. As a result, the overall performance, stiffness degradation, and energy dissipation capacity of masonry shear walls is not well understood. Although current design codes provide equations to predict the shear strength of partially grouted (PG) walls, many of them are empirical based and rely on the behaviour of fully grouted (FG) walls and result in models lacking accuracy. With masonry shear walls used in many seismic regions around the world, there is a strong need for the development of a reliable and efficient design expression applicable to partially grouted masonry shear walls. Artificial neural networks (ANN) have shown great potential in engineering research applications to address highly complex problems and predict nonlinear relationships. By providing an ANN with a set of data of multiple inputs and a corresponding output, it can be trained to determine the weighted effect of each input parameter. This paper presents the development of an ANN analysis model for the shear strength of masonry walls, using a compiled experimental database of PG wall specimens. The effect of previously unaccounted parameters, such as size effects, in code-based approaches is discussed, as well as the influence of different types of ANN analysis options and input size on the model predictions. A sensitivity analysis is performed to evaluate the ANN model and gain insight for future research based on its predictions. The ANN model results are compared against leading design codes in North America (CSA S304 and TMS-402) to predict the shear strength of PG walls.