The minimum variance control law for a cascade control system is derived fo
r two cases: the primary process has stable and unstable inverses. The expr
ession for the primary output variable under the minimum variance cascade c
ontrol is shown to be feed-back-invariant. An estimate of the minimum achie
vable variance in a cascade control loop can be obtained via multivariate t
ime-series modeling of the primary and secondary measurements collected und
er normal operation. The utility of the proposed performance assessment met
hod is demonstrated using simulated and industrial data.