We have demonstrated that a wide variety of pentacoordinate silicon anions (siliconates) should be stable and can be prepared by combining the predictive powers of MNDO and ab initio computational methods and the flowing afterglow (FA) experimental technique. MNDO has been used to compute the anion affinities of 91 siliconates; all but five of these are predicted to be stable with respect to the loss of an anion. Twenty-four siliconates, most of them previously unreported, have been prepared and studied in the FA. The MNDO predictions were, in general, consistent with the experimental results and with trends previously reported by Corriu and co-workers, but in some cases they were found deficient. For example, MNDO tends to underestimate the stability of fluorine-containing siliconates. In these cases, we have carried out ab intio computations and found these to be consistent with both the experimental studies reported here and the Corriu trends.