A relatively recent application area for Virtual Reality (VR) systems is sports training and user performance assessment. One of these applications is eye-hand coordination training systems (EHCTSs). Previous research identified that VR-based training systems have great potential for EHCTSs. While previous work investigated 3D targets on a 2D plane, here we aim to study full 3D movements and extend the application of throughput analysis to EHCTSs. We conducted two user studies to investigate how user performance is affected by different target arrangements, feedback conditions, and handedness in VR-based EHCTSs. In the first study, we explored handedness as well as vertical and horizontal target arrangements, and showed that user performance increases with the dominant hand and a vertical target plane. In the second study, we investigated different combinations of visual and haptic feedback and how they affect user performance with different target and cursor sizes. Results illustrate that haptic feedback did not increase user performance when it is added to visual feedback. Our results inform the creation of better EHCTSs with mid-air VR systems.