Methods and apparatus to use predicted actions in VR environments are disclosed. An example method includes predicting a predicted time of a predicted virtual contact of a virtual reality controller with a virtual musical instrument, determining, based on at least one parameter of the predicted virtual contact, a characteristic of a virtual sound the musical instrument would make in response to the virtual contact, and initiating producing the sound before the predicted time of the virtual contact of the controller with the musical instrument.