We set up a library of virtual fixtures with both haptic and graphic properties and behaviors. For a given task, Virtual Fixture Assembly Language (VFAL) could be used to construct various virtual fixture series, with graphic and force guidance rules, making the low-level haptic and graphic rendering details transparent to the developers. An experiment evaluated the application of virtual fixtures as an aid for guiding a user in a path navigation task. The task was performed with or without force field guidance of virtual fixtures, and then transferred to the condition with no virtual fixtures. Results showed significant learning and transfer effects measured by performance time and path length. However, training using virtual fixtures with force guidance had comparable results to training with graphic only fixtures representing the path. Results are discussed in terms of motor learning theory, future work and applications for the design of better VR training environments.