Health and usage monitoring systems (HUMS) are being incorporated into an increasing number of applications, e. g. monitoring safety critical components in civil, aerospace, mechanical and other structures. A good example is the use of HUMS in monitoring transmission and drive train components on rotary-wing aircraft. These transmission HUMS have enjoyed success in predicting the deterioration of components, however, current system implementations rely on high-bandwidth hardwired sensors and significant data processing capability to perform feature extraction and classification, limiting the locations where they can be installed. To extend HUMS capability into new application areas, such as wind turbine blades or helicopter rotor head components, and other applications impossible to hard wire, the functionality of HUMS needs to be implemented within a wireless sensor network (WSN). The power, processing and packaging constraints of a WSN present many challenges. This paper initially considers the performance requirements of a conventional wired HUMS and contrasts this with that available from state-of the art WSN components. Technical issues related to power supply, sensor technologies, signal conditioning, damage detection and prognostic algorithms for low power microprocessors, robustness and data integrity on wireless radio are discussed. The paper further considers different approaches reported in the literature to overcome system limitations, such as the use of intelligent sensor nodes which perform local signal processing and transmit only a reduced dataset. Finally, simple statistical measures are executed on a low power microcontroller to demonstrate the potential of such devices for damage diagnostics.