An estimation algorithm related to the kinetics terms of the main uncertainties present in a continuous stirred tank bioreactor, is developed. This algorithm is based on a proportional-integral reduced order uncertainty observer. With the estimate generated of the uncertain term, an input-output linearizing feedback control is designed which provides robust regulation model uncertainties, noisy measurements and sustained disturbances. This control strategy could be represented as PI2 controller where new tuning rules of the controller gains are given in terms of the estimation and closed-loop time constants. The performance of the estimation algorithm and the corresponding closed-loop behaviour of the system are compared to a standard PI controller and they are illustrated by means of numerical simulations.