Currently, the operation of a number of sensors is based on the measurement of the optical path length of a micro-Fabry—Pérot cavity. In this paper we consider the impact of manufacturing variations on the performance of such sensors, and discuss designs that minimize the impact of these variations. In particular, we consider how random variations in thickness of the cavity mirrors influence the accuracy with which the gap can be measured. We find that an optimum combination of initial gap and mechanical travel of the cavity that gives the least variation in response curve exists for a given mirror design. This approach determines the device design that maximizes the manufacturing yield of Fabry—Pérot sensors with a specified level of performance.