This paper presents a method for estimation of parameters of a second order plus dead time process model. The estimation procedure is based on transforming the closed loop data into the Fourier frequency domain. Then, using a fairly pedestrian approach the transformed data is used with a linear least-squares estimator to obtain the estimated values. In contrast to former least-squares estimation methods, no iterations or search procedures are implied. This is achieved by using the PID tuning practice that regards self-regulating processes as a first or second order dynamic. Thus the resulted method is attractive for self tuning controllers and related process control applications. Moreover, since closed-loop estimation is adopted, the model parameters are less sensitive to disturbances and measurement noise occurring during the test stage. However, this requires a knowledge of the controller transfer function and settings. The proposed method is applicable to both under- and overdamped closed loops and is not limited only to step test signals. Since the method is simple, its results can be compared against closed-loop reaction curve methods. This is done using a comparative basis from the literature.