An adequate conceptual definition of the geoid is essential for the unambiguous combination of satellite tracking data, satellite al‐timetry, and surface gravity measurements to obtain sea surface topography. The factors influencing the selection of a particular level surface of the earth's gravity field include the purpose(s) for which the geoid is to be used at the 5‐cm level, and the types of data to be used in achieving these objectives. The principal reasons for high precision determinations of the shape of the geoid are: the determination of sea surface topography for applications in oceanography; and the unification of leveling datums with a resolution equivalent to that of first order geodetic leveling. A conceptual definition of the geoid acceptable to oceanographers would be: The geoid for a selected epoch of measurement is that level surface of the earth's gravity field in relation to which the average non‐tidal (or quasi‐stationary) sea surface topography is zero as sampled globally in ocean regions. In the geodetic context, it would be convenient, though not essential, to modify this definition in such a way that the global sea surface topography had zero mean as sampled for evaluations of the geodetic boundary value problem. In either case, a basis exists for unifying all leveling datums serving areas in excess of 106km2, using either gravity anomaly data for the regions or precise determinations of position at first order bench marks. Unfavorable signal‐to‐noise ratios can pose problems when dealing with datums serving smaller areas. Elevation and gravity data banks must be correctly referenced to leveling datums prior to use in sea surface topography determinations. A recent attempt to upgrade the Australian gravity anomaly data bank indicates that all current data banks of this type are inadequate for the task. It is unlikely that time variations in the radial position of the geoid as conceptually defined above, will exceed ±5 cm per century, provided the rate of earth expansion was less than 1 part in 1010yr‐land there is no dramatic change in the present rate of secular change in Mean Sea Level.