The bioleaching of a cobaltiferous pyrite was carried out at 20% solids in a continuous laboratory-scale unit composed of four stirred reactors arranged in cascade. The gas balance method was used in order to follow the O-2 and CO2 consumptions throughout the experiment. The dissolved oxygen concentration below which oxygen availability was limiting was 1.2 mg. l(-1). In the absence of oxygen limiting conditions, the pyrite oxidation was not directly linked to the bacterial growth metabolism. However, the quantity of biomass formed in the first steps of bioleaching influenced the overall efficiency of cobalt recovery. It was found that high agitation rates could affect bacterial productivity. The use of ammonium rather than urea improved the bioleaching efficiency by favouring bacterial attachment to the solid substrate. The oxygen requirements for pyrite oxidation were estimated on average to be 0.89 kg of O-2 per kg of pyrite oxidised. The study was also used to evaluate the organic matter distribution and to propose a mathematical model of the bioleaching process. This model, which was calibrated with experimental data, demonstrated a good level of efficiency in its prediction of cobalt recovery for different cascade configurations.