This paper presents the development of a multivariable nonlinear adaptive controller for perfusion bioreactors, and its simulated behavior on a model that has been identified from experimental data. A contribution to the bioprocess model is also proposed, which is supported by experimental observations. The proposed control strategy is a multivariable approach to regulate the biomass and substrate concentrations using the fresh medium addition and direct bleeding streams as the manipulated variables. Level control would be ensured by a proportional integral (PI) control loop, using either the perfusion flow (draining flow that retains the cells in the reactor) or a nutrient-free phosphate-buffered saline (PBS) solution flow added to the reactor. The flow that is used for level control determines the operation mode of the reactor, being perfusion or chemostat. This paper presents the controller design, switching considerations between the operation modes, and parameter tuning guidelines. The controller tunings are essentially obtained by pole placement, based on linearization of the closed-loop dynamics. Simulation results prove the technique to be rather efficient, while the transitions between the operation modes are smooth and without risks.