This addendum paper presents additional material on an output feedback eigenstructure assignment approach derived in Clarke et al. (2003). There, a two-stage process is followed in which a subset of the desired eigenvectors is assigned in the first stage, followed by a dual set in the second stage. Depending upon the number of available degrees of freedom, and the manner in which they are distributed between the stages, it is possible that not all are used. A technique to recover these unused degrees of freedom in a final and independent stage is presented here. The technique, known as retro-assignment, is illustrated using an extension to the original numerical example.