It is of general interest to explore the limits of the molecular replacement method. Described here are the specifics of a successful molecular replacement structure determination in a difficult case: the X-ray crystal structure of a Trypanosoma brucei phosphoglycerate kinase (PGK) ternary complex. This ternary complex crystallizes with four 45 kDa subunits in the asymmetric unit, whereas the available search models were all monomers consisting of two distinct domains. Initial molecular replacement attempts using complete subunits were unsuccessful. Attributing this failure to a presumed change in the relative orientations of the N- and C-terminal phosphoglycerate kinase domains, a second attempt was made using each domain as an independent search model. In this way, two N-terminal and then two C-terminal domains could be oriented and positioned in the unit cell. On the basis of this result, a new search model containing both domains in the correct mutual orientation was created and used to identify the two remaining phosphoglycerate kinase subunits. The ability to successfully orient and position an N-terminal domain containing 8.4% of the scattering mass in the asymmetric unit was the key to this structure determination. Further investigations show that a truncated version of this search model containing 5.8% of the scattering mass would have been sufficient for this purpose. A retrospective analysis suggests that the effectiveness of this probe is enhanced by structural conservation, retained temperature factors and a disparity in the degree of order among the various subunits in the T. brucei PGK asymmetric unit. Based on these observations, 231 well ordered Calpha atoms were selected from a single refined T. brucei PGK subunit and it was demonstrated that this collection of atoms, representing just 1.6% of the scattering mass, could be correctly oriented and positioned in the unit cell.