Match Index represents a quality characteristics of 3D structure alignment. Each 3D alignment is described by several quantities, from which the most important are alignment length and RMSD. In simple words, a quality characteristics provides a measure of the alignment's quality in a single number.

Although many definitions for quality characteristics may be suggested, SSM employs the one, match index, introduced in Ref. [1]: 

                  1 + Nalgn
    MI = ----------------------------
         (1 + RMSD/R0)(1 + min(N1,N2))
In this expression, Nalgn is the number of aligned (matched) residues, RMSD - achieved r.m.s.d, N1 and N2 is the number of residues in 1st and 2nd structures. R0 is the calibration parameter, measuring the significance of RMSD against Nalgn.

One can easily see that MI changes from 0 to 1 monotonically with changing both Nalgn and RMSD. The lower limit (close to 0) is reached at large RMSD and/or low algn; the upper limit, equal to 1 is reached at perfect match with RMSD=0 and Nalgn=min(N1,N2). The larger R0, the less sensitive is MI to the variation of RMSD.

It is clear that higher MI corresponds to longer alignment at same RMSD. It is therefore possible to say that higher MI should indicate a better quality of alignment. In the course of 3D Ca-alignment, SSM tries to maximize the match index MI at R0=1.5 Å, the latter has been chosen empirically.

3D alignment is usually a compromise between alignment length and RMSD. With the exception for the case of highly similar structures, all or almost all residues of which are normally aligned, alignment algorithm may offer longer alignment lengths at the expense of higher RMSDs or vice versa. That, in fact, is a question of calibration, i.e. tuning the algorithm's parameters in favor of the alignment length or RMSD. Given this situation, match index represents a characteristics which should be less sensitive (ideally insensitive) to the calibration parameters, provided they are in a sensible range. Indeed, a clear finding of our comparative analysis is that different 3D alignment algorithms, available in the Web, show strikingly similar match indexes, while the difference in alignment length and RMSD may be significant.

------------------------------------
[1] G.J. Kleywegt and A. Jones (1994). super position. CCP4/ESF-EACBM Newsletter on Protein Crystallography 31, November 1994, pp. 9-14.