NUMERICAL weather prediction has been operational since 1955 (Thompson, 1961;Fawcett, 1962). However, insufficient data and computing power have precluded a similar capability for the ocean despite the potential for numerous military and civilian applications, e.g.: antisubmarine warfare, tactical planning, optimum-track ship routing, search and rescue, long-range weather and climate prediction, sea-ice prediction, fisheries planning, design and protection of underwater structures such as oil figs, and prediction of pollutant dispersion. Obstacles to prediction of ocean circulation arise because, historically, the ocean has been much more difficult to observe. In addition, the spatial scale for meandering ocean currents and eddies (the oceanic "weather") is an order of magnitude smaller than that for major weather systems and the meandering of the jet stream. It is only now that global-scale, eddy-resolving ocean prediction is at the threshold of feasibility. Four key technical or technological advances have made this possible.1. Class VII supercomputers, like the Cray Y-MP8/8128, are the first capable of executing truly eddy-resolving global and basin-scale ocean models using the most efficient ocean model design available.2. Global and basin-scale ocean models and data-assimilation techniques are being developed that can effectively utilize the expected data types. These models are designed for efficient execution on present and future supercomputers, including massively parallel machines. In the system under development by the U.S. Navy, mixed-layer models with about 20 levels in the upper 400 m are coupled to eddy-resolving ocean circulation models with about 6 layers in the vertical. This is a key strategy in developing the system, because it reduces the computational requirements by two orders of magnitude.3. Satellites provide useful data with global coverage and adequate space-time resolution and accuracy. This includes oceanic data that the models can assimilate and also includes atmospheric forcing functions. Sea-surface height (SSH) from satellite altimetry is the single most promising source of oceanic data for operational ocean prediction, because it is strongly related to subsurface thermal structure and a major component of oceanic surface currents. Other notable satellite data sources include infrared radiometers (IR), scatterometers, multichannel microwave radiometers, and ocean-color imagers. These provide information such as the location of oceanic fronts and eddies, sea-surface temperature, and surface winds and heat fluxes.4. Another advance is improved data communications via satellite, especially shore to ship.These topics are discussed in more detail by Hurlburt (1984 and Malanotte-Rizzoli and Hurlburt (1987). Navy interest includes depicting the present state of the ocean, a process sometimes called "nowcasting," and forecasting its future state. Skill in doing this is measured by improvement over oceanic climatology or (more common in forecasting) the improvement over a forecast of pers...