Traditionally, when some new concept or equipment comes along, one first makes a rough judgment based on a philosophical analysis. If this looks sensible the next step is usually to put the new device to sea and evaluate it there. The early philosophy, the users' reports and a simple economic evaluation usually lead to a sufficiently sound decision. A collision-avoidanoe radar of the new generation presents a substantial philosophical hurdle both to the marine superintendent ashore and to the man at sea, and it represents a fairly substantial capital expenditure. Deciding what is the ‘best buy’ becomes a complex matter for which traditional selection methods may be inadequate. Like others, in our efforts to provide the most useful installation at economic cost, we have attempted sea-trials of some of the new systems. It has proved most difficult to reach objective conclusions, mainly because normal trading does not permit a free choice of encounter situations or manoeuvres, and there is no ‘control’ experiment. In addition, user reports were sometimes dominated by lost confidence due to technical teething troubles. We therefore felt a need to break the problem down into its elements and to attempt a controlled and objective analysis of these elements.