The Astrom-Hagglund autotuner is extended to multivariable systems. The proposed multivariable automatic tuning performs the identification-design procedure in a sequential manner. Its advantages are discussed, as well as the process characteristics resulted from sequential design. One important feature is that the sequential design often leads to underdamped process characteristics. Given these characteristics, potential problems in the Ziegler-Nichols tuning are discussed and a modification is made. The sequential identification and the modified Ziegler-Nichols controller design method form the basic structure for the multivariable autotuner. The properties of con vergence, the implications of tuning sequence, and the indications of undesirable pairing for the autotuner are explored. Consequently, a multivariable autotuning procedure is proposed. Two nonlinear distillation examples and a 3 X 3 linear system are used to illustrate the effectiveness of the autotuner. Simulation results show that good performance can be obtained with minimal engineering effort and the autotuner works well for very difficult processes in a very simple way.