It is well known that ship performance is degraded in rough weather. In moderately severe conditions every task on board the ship will take longer than it does in calm weather. The passengers and crew may be seasick and they will have to take more care when moving around the ship to avoid injury. The physical performance of mechanical and electrical systems may also be reduced if the ship motions are severe. Cargo may be damaged and this may also be considered as a reduction in performance. In extreme conditions, the ship may capsize or founder and this is, of course, the ultimate loss of performance. The aim of research in seakeeping is, or should be, to develop techniques of predicting the degradation of performance a ship will experience in rough weather. If this can be achieved it will enable the ship designer to eliminate unsatisfactory and unsafe ships at an early stage in the design process. The paper begins by summarizing the standard techniques for predicting the motions a ship will experience in rough weather. These are based on well-known strip theories coupled with superposition techniques for inferring the statistics of the motions in irregular waves from the regular wave transfer functions. These techniques are quite well validated for conventional monohull ship motions in the vertical plane but predictions of motions in the lateral plane are rather less reliable and further work is needed in this area. Accurate predictions of the irregular motions in realistic sea states are only one step in achieving the ultimate goal of predicting the rough-weather performance of new designs of ships. We also need to estimate the maximum permissible levels of the motions (usually known as seakeeping criteria). There are, in principle, no universally applicable criteria. They depend on the activities within the ship while it is engaged in a given mission and also on the type of equipment used. We therefore need to quantify the way in which the performance of the various sub systems or tasks which make up the ship mission degrade in rough weather. Ideally the first step should be to examine each task (or at least a representative selection of the important tasks) and to study the way in which the performance degrades. This will allow the motions (or other rough-weather phenomena) which are important for that particular task to be identified. It is then necessary to determine the maximum permissible levels of these motions by simulation, questionnaires, trials or even guesswork. Finally there is a need to develop universally acceptable measures of performance. Ideally these might include elements of the time required to complete a task (in relation to the time required in calm weather) as well as the quality of the result.