The difference between Monte Carlo Treatment Planning (MCTP) based on the assumption of 'water-like' tissues with densities obtained from CT procedures, or on tissue compositions derived from CT-determined densities, have been investigated. Stopping powers and electron fluences have been calculated for a range of media and body tissues for 6 MV photon beams, including changes in their physical data (density and stopping powers). These quantities have been used to determine absorbed doses using cavity theory. It is emphasized that tissue compositions given in ICRU or ICRP reports should not be given the standing of physical constants as they correspond to average values obtained for a limited number of human-body samples. It has been shown that mass stopping-power ratios to water are more dependent on patient-to-patient composition differences, and therefore on their mean excitation energies (I-values), than on mass density. Electron fluence in different media are also more dependent on media composition (and their I-values) than on density. However, as a consequence of the balance between fluence and stopping powers, doses calculated from their product are more constant than what the independent stopping powers and fluence variations suggest. Additionally, cancelations in dose ratios minimize the differences between the 'water-like' and 'tissue' approaches, yielding practically identical results except for bone, and to a lesser extent for adipose tissue. A priori, changing from one approach to another does not seem to be justified considering the large number of approximations and uncertainties involved throughout the treatment planning tissue segmentation and dose calculation procedures. The key issue continues to be the composition of tissues and their I-values, and as these cannot be obtained for individual patients, whatever approach is selected does not lead to significant differences from a water reference dose, the maximum of these being of the order of 5% for bone tissues. Considering, however, current developments in advanced dose calculation methods, planning in terms of dose-to-tissue should be the preferred choice, under the expectancy that progress in the field will gradually improve some of the crude approximations included in MCTP and numerical transport methods. The small differences obtained also show that a retrospective conversion from dose-to-tissue to dose-to-water, based on a widely used approach, would mostly increase the final uncertainty of the treatment planning process. It is demonstrated that, due to the difference between electron fluence distributions in water and in body tissues, the conversion requires an additional fluence correction that has so far been neglected. An improved expression for the conversion and data for the fluence correction factor are provided. These will be necessary even in a dose-to-tissue environment, for the normalization of the treatment plan to the reference dosimetry of the treatment unit, always calibrated in terms of absorbed dose to water.