For a natural interaction, people immersed within a virtual environment (like a CAVE system) use multimodal input devices (i.e. pointing devices, haptic devices, 3D mouse, infrared markers and so on). In the case of physically impaired people who are limited in their ability of moving their hands, it is necessary to use other special input devices in order to be able to perform a natural interaction. For the inference of their preference or interests regarding the surrounding environment, it is possible to take in consideration the movements of their eyes or head. Based on the analysis of eye movements, an assistive high level eye tracking interface can be designed to find the intentions of the users. A natural interaction can also be performed at some extent using head movements. This work is a compared study regarding the promptness of selection between two interaction interfaces, one based on head tracking and the other based on eye tracking. Several experiments have been conducted in order to obtain a selection speed ratio during the process of selecting virtual objects. This parameter is useful in the evaluation of promptness or ergonomics of a certain selection method, provided that eyes focus almost instantly on the objects of interest, long before a selection is completed with any other kind of interaction device (i.e. mouse, pointing wand, infrared markers). For the tests, the tracking of eyes and head movements has been performed with a high speed and highly accurate head mounted eye tracker and a 6 DoF magnetic sensor attached to the head. Direction of gaze is considered with respect to the orientation of head, thus users are free to turn around or move freely during the experiments. The interaction interface based on eye tracking allows the users to make selections just by gazing at objects, while the head tracking method forces the users to turn their heads towards the objects they want to be selected.