Theoretically a crystal may contain both complexed and uncomplexed molecules simultaneously in the same lattice. Since we seldom screen for such possibilities, such occurrences are only rarely reported. Here we propose that stoichiometry should be one of the parameters to be screened in the crystallization of macromolecular complexes. By allowing for non-biologically significant stoichiometries, we may increase the chances of crystallizing a macromolecular complex and of selecting arrangements which crystallize better or yield more ordered crystals. Although biological forces tend to be stronger than lattice-building interactions, in the crystal the latter will dominate numerically. By allowing for a varied stoichiometry we permit a wider selection of lattice-building contacts and increase the probability of crystallization. From these theoretical considerations we have developed methodology compatible with classical solubility screening and other well-established crystallization principles. We discuss this technique, stoichiometric variation screening (SVS), as part of a multicomponent system for the enhancement of crystallization of macromolecular complexes. We present this technique as an extension of reverse screening and illustrate the complementarity in the methodology. We present two examples of the use of SVS: the complexes between an immunoglobulin Fab fragment and two bacterial proteins, namely the D domain of protein A from Staphylococcus aureus (SpA) and a single domain of protein L from Peptostreptococcus magnus (PpL). In the first example there are 3 Fab molecules and only 2 SpA D domains (domD) (2 complexed and I unliganded Fab), in the second 2 Fabs and only I PpL domain (I complexed and I unliganded Fab). SVS has the added and unique advantage that in the same crystal we have information on both the unliganded and complexed states under precisely identical conditions: one structure, two answers. Together with a combinatorial method for complex crystallization based on immunoglobulin-binding domains, it may enhance the probability of crystallization by well over a factor of ten.