For applications where accurate speed of sound estimates are needed, computation errors arising from direct use of equations of state, or from use of mixing rules based on pure component property data are too large. In this work we translate isentropic compressibility estimates for mixtures at specific compositions by adding the molar average error of the predicted isentropic compressibility for pure constituents to the isentropic compressibility of the mixture predicted using the SAFT-VR-Mie EOS. The translated isentropic compressibility values and mixture densities predicted by the EOS are then used to compute the speed of sound in mixtures. This procedure leads to better qualitative descriptions of the dependence of speed of sound in 1 - n-alkanol + n-alkane mixtures vis-A-vis current practice. The rms errors obtained are, on average, substantially lower than other methods that only make use of pure component data as inputs, and they are competitive with methods such as the Extended Real Association model (ERAS) which require composition specific mixture properties. The success of this computational approach may suggest that the SAFT-VR-Mie EOS can be used as a basis for speed of sound calculations for mixtures more generally. For n-alkanol + n-alkane mixtures, refinement of the repulsive terms in the SAFT-VR-Mie EOS is warranted because the ideal mixing rule out performs this computational approach in some cases. (C) 2009 Elsevier B.V. All rights reserved.