The goal of this study is to identify on-line a dynamic model of biodegradation in soil to optimize biological processes by the implementation of advanced control strategies. Namely, the theory of adaptive systems leading to the development of software sensors is investigated to deal with non-measured state variables and unknown parameters involved in the hydrodynamic and biokinetic models. A numerical model capable of describing water flow and biodegradation in porous media is developed and calibrated using laboratory experiments. First, an estimator based on the suction measurement IS ble to estimate on-line the hydraulic conductivity in steady-state but faces oscillation problems in unsteady-state. Second, using on-line carbon dioxide measurements, an asymptotic observer allows the estimation of hexadecane depletion with a relative error below 5% on the overall time trajectory. This result is significant, as it indicates the potential for real-time monitoring of the performance of treatment bioprocesses in unsaturated soils. Finally, these results demonstrate that the theory of adaptive systems could provide a valuable tool for implementation of advanced control strategies in soil bioremediation processes.