The algorithms available today that use dipolar coupling data for macromolecular structure determination require the independent determination of two parameters, Da PQ and R. Methods exist for obtaining these parameters when the set of dipolar couplings available is large and the orientations of the interatomic vectors on which they report is isotropically distributed. These methods are less satisfactory when the set is small and anisotropic. Described here is a maximum likelihood method that extracts accurate values for Da PQ and R from small, anisotropic data sets. Also demonstrated is a procedure for estimating the errors associated with the values of Da PQ and R obtained and for incorporating these errors into refinement protocols.