Abstract A procedure is described for computing sedimentation coefficient distributions from the time derivative of the sedimentation velocity concentration profile. Use of the time derivative, ( ∂c ∂t ) r , instead of the radial derivative, ( ∂c ∂r ) t , is desirable because it is independent of time-invariant contributions to the optical baseline. Slowly varying baseline changes also are significantly reduced. An apparent sedimentation coefficient distribution (i.e., uncorrected for the effects of diffusion), g ∗(s) , can be calculated from ( ∂c ∂t ) r as g ∗(s) t = ∂c ∂t corr 1 c o ω 2t 2 1n( r m r ) r r m 2 where s is the sedimentation coefficient, ω is the angular velocity of the rotor, c 0 is the initial concentration, r is the radius, r m is the radius of the meniscus, and t is time. An iterative procedure is presented for computing g ∗(s) t by taking into account the contribution to ( ∂c ∂t ) r from the plateau region to give ( ∂c ∂t ) corr . Values of g ∗(s) t obtained this way are identical to those of g ∗(s) calculated from the radial derivative to within the roundoff error of the computations. Use of ( ∂c ∂t ) r , instead of ( ∂c ∂r ) t , results in a significant increase (>10-fold) in the signal-to-noise ratio of data obtained from both the uv photoelectric scanner and Rayleigh optical systems of the analytical ultracentrifuge. The use of ( ∂c ∂t ) r to compute apparent sedimentation coefficient distributions for purposes of boundary analysis is exemplified with an antigen-antibody system.