This work deals with the identification of transfer function model of an unstable bioreactor by a subspace-based identification method. The data set is generated from the simulated closed loop system by giving a random signal at the reference signal. The linear system matrices are estimated through the Multivariable Output Numerical algorithms for Subspace State Space System Identification (MON4SID) algorithm. The delay is estimated directly from the input–output data by the impulse response method. The identified First Order Plus Time Delay (FOPTD) model is compared with the locally linearized transfer function model derived around the unstable operating point. A Proportional Integral Derivative (PID) controller is designed for the identified model using the equating coefficient method. The closed loop servo and regulatory responses show that the model identified by MON4SID is similar to the linearized model and the optimization method. © 2015, Copyright © Taylor & Francis Group, LLC.