<p>Electrochemical jet machining (EJM) is an emerging electrochemical machining variant, demonstrating numerous advantages in surface modification. While EJM has been extensively applied to planar surfaces, research on machining curved surfaces remains limited. This study investigates the machining mechanisms and shape prediction of EJM on curved surfaces through simulations and experiments. A three-dimensional multiphysics finite element model is developed to clarify the effects of workpiece curvature and jet angle on jet shape, electric field, and material removal patterns. The results indicate that workpiece curvature and jet angle variations affect the jet shape and machining gap distribution, altering the electric field distribution and resulting in different removal patterns. A decrease in the jet angle leads to deeper material removal. The removal depth for convex surfaces increases with curvature, whereas the opposite trend is obtained for concave surfaces. Additionally, the relationship between removal depth, workpiece curvature, and jet angle was established through systematic simulations. Based on this relationship, an analytical approach is developed for predicting the shape created on curved surfaces by EJM. Shape prediction and machining experiments are conducted for EJM grooving and surface finishing on additively manufactured hemispherical surfaces, taken as the representative curved surface. The predicted shapes align well with the experimental results. These findings provide an in-depth understanding of the physics interactions in EJM of curved surfaces, contributing to the advancement of EJM theory. Moreover, the proposed shape prediction method facilitates the efficient development and optimization of the EJM process for curved surfaces, thereby broadening its potential applications.</p>