The use of a single high-speed sedimentation equilibrium experiment to analyze mixed associating systems is inadequate to determine the association mode even if the molecular weight of one species is known. Simultaneous analysis of the concentration distributions at three equilibrium speeds greatly reduces the lack of uniqueness. Linear least-squares multi-speed fits discriminate between association models in which the molecular weights are assumed. Experiments at a series of initial concentrations as well as rotor speeds further increases the discrimination. An overspeed procedure is proposed. The overspeed time depends only slightly on the sedimenting species molecular weight, but significantly on the frictional ratio. A minimal overspeed time may be estimated.