A stochastic model of oil-palm (Elaeis guineensis Jacq.) root system architecture and development has been developed. This model enabled us to create 3-D numerical models of complete root systems by simulation. The application of a postprocessor software, called "RACINES", to these 3-D numerical models, provided an estimation of some parameters of plant root systems. The objective of this paper is to present oil-palm root characteristics as possible outputs of the application of this RACINES software. The outputs described in this article cover (i) spatial distribution of roots under plantation conditions, (ii) the estimation and distribution of total root biomass, per root type or per soil horizon and (iii) the location and quantification of absorbent surfaces. The computing techniques used were based on voxellization of space and creation of 3-D virtual sceneries exactly reproducing observed planting designs. By comparing the results of observations and simulations for spatial distribution (by trench wall density maps) and root biomasses (by real and virtual sampling) we were able to carry out additional numerical validations of the model.