The core of human immunodeficiency virus type 1 is derived from two precursor polyproteins, Pr55gag and Pr160gag-pol. The Gag precursor can assemble into immature virus-like particles when expressed by itself, while the Gag-Pol precursor lacks particle-forming ability. We have shown previously that the Gag precursor is able to "rescue" the Gag-Pol precursor into virus-like particles when the two polyproteins are expressed in the same cell by using separate simian virus 40-based plasmid expression vectors. To understand this interaction in greater detail, we have made deletion mutations in the capsid-coding regions of Gag- and Gag-Pol-expressing plasmids and assayed for the abilities of these precursors to assemble into virus-like particles. When we tested the abilities of Gag-Pol precursors to be incorporated into particles of Gag by coexpressing the precursors, we found that mutant Gag-Pol precursors lacking a conserved region in retroviral capsid proteins, the major homology region (MHR), were excluded from wild-type Gag particles. Mutant precursors lacking MHR were also less efficient in processing the Gag precursor in trans. These results suggest that the MHR is critical for interactions between Gag and Gag-Pol molecules. In contrast to these results, expression of mutated Gag precursors alone showed that deletions in the capsid region, including those which removed the MHR, reduced the efficiency of particle formation by only 40 to 50%. The mutant particles, however, were clearly lighter than the wild type in sucrose density gradients. These results indicate that the requirements for Gag particle formation differ from the ones essential for efficient incorporation of the Gag-Pol precursor into these particles.