Ray casting is frequently used to point and select distant targets in Virtual Reality (VR) systems. In this work, we evaluate user performance in 3D pointing with two different ray casting versions: infinite ray casting, where the cursor is positioned on the surface of the first object along the ray that said ray points at, and finite ray-casting, where the cursor is attached to the ray at a fixed distance from the controller. Twelve subjects performed a Fitts' law experiment where the targets were placed 1, 2, or 3 meters away from the user. According to the results, subjects were faster and made fewer errors with the infinite ray length. Interestingly, their (effective) pointing throughput was higher when the ray length was constrained. We illustrate the advantages of both methods in immersive VR applications and provide information for practitioners and developers to choose the most appropriate ray-casting-based selection method for VR.