Volume rendering is an emerging technique in the field of seismic data visualization. Only recently commercial software has been made available to the oil industry. Volume rendering is particulary useful for understanding structural and stratigrafhic features present in seismic data in a more natural way than using the tradicional method of interpreting over a set of 2D slices. This work adresses several topics related to the nature of seismic data and to how they interrelate with techniques commonly used in volume redering.Special attention is given to three topics: quantization of acquired seismic amplitudes to 8-bit representations, segmentattion techiniques to highlight seismic events and use of illumination models.The quantatization effects are analyzed by appying different color quantization algorithms adapted to handle seiosmic amplitudes. Uniform quantization is selected due to its simplicity and efficiency. The segmentation of seismic events is studie theorecally and verified experimentally.Several characteristics of seismic data that prevent segmentation frombeing successful are discussed. The significance of shading obtained by applying common illumination models is studie theoretically. The estimation of normal vectors directly from amplitude values is shown to be inadequate. Finally, several basic requisites to be fulfilled by seismic data volume visualization algorithms are identified.