The combined application of astronomic, geodetic, and gravimetric data to computations of the geoid is discussed. The general principle observed is that any adjustment should weight all observations inversely as their variances. Two conditions are imposed: (1) Geoid heights and deflections computed by Stokes' theorem from gravity data must equal those derived by astrogeodetic means. (2) The five harmonicsP1,P11sin λ,P11cos λ,P21sin λ, andP21cos λ must be absent from the adjusted gravity field. The ideal case is discussed, including provision for separate determinations of parameters by independent astronomic methods. Practical modifications are then introduced in turn: treating gravity anomalies as representative of areas; holding geodetic or astronomic observations constant; using a reduced number of astronomic stations; and comparing interpolated points of the astro‐geodetic and gravimetic geoids. The most significant discrepancy from the ideal case of most practical solutions made heretofore appears to be in the weighting of the gravimetric