This study examines the biomechanical effects of idealized multi-joint exoskeleton assistances on hip, knee, and ankle joints. We conducted predictive simulations of walking without assistance and with seven different assistance cases including assistance to each joint, assistance to any two joints, and assistance to all three joints. A 2D musculoskeletal model with 10 degrees of freedom and 18 muscles was used and the OpenSim Moco optimal control solver was employed for all predictive simulations, which aimed to minimize the weighted sum of several objectives including metabolic cost, muscle activation, joint coordinate acceleration, motion tracking, and whole-body center of mass (COM) acceleration. The results showed that all assistance cases changed the joint kinematics of the walking motion to different degrees and for most cases the exoskeleton assistance reduced muscle effort substantially. By comparing with the unassisted case, we found that the two cases with assistance to all three joints and to the hip-ankle joints both provided more than 50% reduction in metabolic cost of transport (COT), followed by assistance to hip-knee and knee-ankle joints with less than 40% reduction. As for the single joint assistance cases, assistance to the hip joint appeared to be the most effective with around 34% reduction in COT, followed by the assistance to the ankle joint with around 22% reduction, whereas the assistance to the knee joint was much less effective (with less than 10%).