Previously we have identified an immunodominant, eight-residue, epitope core sequence (TAAGNVNI) from the 19,000 MW protein of Mycobacterium tuberculosis, which is recognized in the context of multiple H-2 I-A molecules. In this study, the role of residues flanking this T-cell epitope core was examined, using a series of 20 mer analogue peptides in which the native flanking residues were progressively replaced with L-alanine. Analogue peptides were tested for their capacity to stimulate a CD4+ 19,000 MW protein-specific T-cell line, revealing that all but one N-terminal flanking residue could be replaced collectively by alanine without significant loss of stimulatory activity. However, clear H-2-associated differences in the requirement for flanking residues were demonstrated with peptide-specific T-cell hybridomas. In particular, H-2d-derived hybridomas were much more stringent in their requirement for flanking residues than were H-2b hybridomas. All polyalanine-substituted peptides bound I-Ab molecules, with affinities similar to the native unsubstituted peptide. In contrast, significantly reduced binding to I-Ad was observed with several analogue peptides, although without a clear relationship to the degree of substitution. Furthermore, in H-2b mice, neither immunogenicity nor cross-reactivity with the native peptide showed a clear inverse relationship with respect to the degree of alanine substitution. The results presented in this paper indicate that flanking residues can influence T-cell specificity and that these effects may be controlled by major histocompatibility complex (MHC) haplotype.