The research reported in the present paper contemplated the study of an undisturbed block of gneissic residual soil moulded in situ and subjected to contamination by petroleum hydrocarbons. The block was instrumented, monitored and submitted to a horizontal flow of a solution composed of water, benzene and toluene to assess the compounds' mobility and decay. For the experiment, 11 monitoring points were assessed for about 3 months. The results showed retention of the compounds and the presence of microbial activity suggesting the occurrence of biodegradation. This article presents the results of a numerical study conducted to simulate the experiment performed with the soil block and assess the model parameters’ sensitivity. The multiplicative Kinetic Model Monod was selected to represent the contaminant biodegradation process. The kinetic parameters were calibrated based on one of the monitoring points previously selected. Results indicated increasing discrepancies with the experimental data as the distance between the sample points and the source increased. The monitoring point presenting the best estimates was selected to calibrate the parameters and assess the sensitivity of the kinetic and transport parameters used in the model. The parameters with the highest sensitivity were bacterial maximum specific growth and death rate, as well as the half-saturation coefficient of benzene. The least sensitive parameters were bacteria concentration and dispersivity.