SESAMI, a comprehensive program for the elucidation of the structure of complex compounds of carbon, incorporates a structure reduction-based structure genearator (COCOA). Observed limitations with this program in the solution of higher molecular weight unknowns prompted the development of a structure generator (HOUDINI) which embodies a new concept, convergent structure generation. A comparison of the performance of COCOA-based and HOUDINI-based SESAMI using a set of complex, naturally occurring compounds as a test set of unknowns revealed faster execution times and more efficient processing of ambiguous structural information for the latter.