Modern radar systems are required to perform a multitude of functions including highly accurate detection, parameter measurement, classification and tracking of targets over long distances. These targets may have low effective visibility and exist in a hostile environment of noise and interferences. Significant improvements in traditional monostatic radar require brute-force approaches such as larger antennas and power amplifiers, which are impractical and expensive. Multistatic radar, comprising a system of multiple, spatially separated transmitters and receivers, is one promising solution to this problem. This thesis concerns the design, development and construction of such a radar at low cost, in particular where each dispersed component of the system is mutually coherent and networked to allow cooperative operation and the joint processing of all received signals. The statistical theory of multistatic detection is analysed and processing algorithms are developed for implementation in the system. Models for the predicted coverage of the radar are developed, and illustrations of the system instrument function axe presented based on the derivation of the ambiguity function for a range of topologies and modes of operation. The requirements for obtaining spatial coherency across the system are considered, and methods of fulfilling these requirements at low cost are devised. A complete design strategy for the radar is developed, based on the use of commercial components and open architecture interfaces. The development of each major subsystem is explained, and the construction of the multistatic radar completed. Finally, the system is tested and calibrated, and some initial experiments are performed in order to determine its performance and demonstrate the advantages of this type of radar.