In modal identification, the use of different actuator and sensor locations results in different frequency response functions (FRFs) and consequently affects the modes shown on the FRF plots. This paper presents a method of placing sensors and actuators to obtain reliable measured data for estimating the modal parameters. The method is based on proposed controllability and observability measures for second-order linear systems, which are suitable for identification purposes and have a physical interpretation on the FRF plots. These measures, combined with an objective criterion, provide a method for ranking the effectiveness of alternative actuator and sensor distributions and hence a rational basis for choosing their locations. The method is applied successfully to the selection of actuator and sensor locations for a set of target modes for the structural characterisation of a simulated system. Finally, an iterative procedure for performing modal surveys is proposed.