Dynamic matrix control (DMC) is based on two assumptions which limit t
he feedback performance of the algorithm. The first assumption is that
a stable step response model can be used to represent the plant. The
second assumption is that the difference between the measured and the
predicted output can be modeled as a step disturbance acting on the ou
tput. These assumptions lead to the following limitations: 1. Good per
formance may require an excessive number of step response coefficients
. 2. Poor performance may be observed for disturbances affecting the p
lant inputs. 3. Poor robust performance may be observed for multivaria
ble plants with strong interactions. Limitations 1 and 2 apply when th
e plant's open-loop time constant is much larger than the desired clos
ed-loop time constant. Limitation 3 is caused by gain uncertianty on t
he inputs. In this paper we separate the DMC algorithm into a predicto
r and an optimizer. This enables us to highlight the DMC limitations a
nd to suggest how they can be avoided. We demonstrate that a new model
predictive control (MPC) algorithm, which includes an observer, does
not suffer from the listed limitations.