In this paper, an integrated identification and control procedure is studied. This integrated procedure seeks to find a high performance controller for real world systems. The procedure, which is inherently iterative, involves three steps in each iteration: (1) closed loop identification; (2) system model extraction from the closed loop experiment data; (3) controller design. The algorithm proposed in this paper uses weighted closed loop identification for deducing a model. The weight used for identification is obtained from the control design step and it provides a measure to evaluate the relative importance of each output channel in the closed loop behavior. Hence the weighted identification can capture models which are good for control design so as to achieve better closed loop performance. The procedure is demonstrated by controlling a smart structure under development at Purdue.