The idea of using variable measurement structures for chemical process control is an attractive one but in the past it has encountered significant computational and implementational difficulties. In the present work an on-line sequential analysis of the measurement equations is presented, which allows for a fast and robust updating of the best measurement structure for the control of several processing systems where updating the measurements' configuration would result in a robust control behaviour. The present paper develops the theoretical foundations for the concept of the variable measurement structures, and implements it for systems operating in a pseudo-steady state or dynamic environment. Furthermore, a strategy is developed for the identification of the optimum measurement structure in the presence of disturbances. Finally the concept of variable measurement structures is discussed with reference to the control of two processing units of interest to the chemical industry ; a distillation column and a tubular chemical reactor.