Interdisciplinary applications exposed new interesting, but difficult problems in the field of measurement. Such applications put emphasis on the automation of the measurement process on the whole. The analysis of how the experiments are planned, controlled, evaluated, etc. brought into focus the question of how well the expert knowledge related to the measurement can be represented and utilised in the development of autonomous intelligent measurement systems . Methods developed in the field of Artificial Intelligence (AI) lend themselves naturally for such applications. The present paper proposes so called constraints to be used as the knowledge representation tool in modelling measurements. When a deeper model of the phenomena underlying the measurement processes is needed, applying constraints as knowledge representation is potentially fruitful, considering how well they express the dependencies between different subsystems and state variables of the physical systems. Authors review measurement technical problems suited for the constraint based knowledge representation, analyse the requirements of the actual constraint satisfaction problems and evaluate the known methodology from that perspective.