The NarI sequence is known to be the strongest mutation hot spot for induced frameshift mutagenesis. Indeed, a single N-2-acetylaminofluorene (AAF) adduct induces −2 frameshift mutations (5′-GGCGAAFCC → 5′-GGCC) more than 107-fold over background mutagenesis in Escherichia coli. The mechanism of induction of the frameshift mutation involves a two nucleotide primer-template misalignment event during replication of the adduct-containing sequence. The slipped mutagenic intermediate (SMI) that is thus formed is strongly stabilised by the AAF residue. In order to understand the origin of the extreme susceptibility of this sequence to frameshift mutagenesis, we analysed AAF-induced mutagenesis at sequences 5′-NaGCGAAFCNb-3′ containing the core dinucleotide GCGC repeat present in the NarI sequence flanked by variable nucleotides Na and Nb. The nature of nucleotide Nb was found to strongly modulate the frequency of induced −2 frameshift mutagenesis (up to 30 to 50-fold), while little if any effect could be attributed to nucleotide Na. The induction of −2 frameshifts, regardless of nucleotides Na and Nb, was found to be SOS-inducible but umuDC-independent as previously found for the authentic NarI sequence. The NarI sequence (GGCGCC) and sequence TGCGCA (Na = T, Nb = A) were found to be equally “hot” for −2 frameshift mutation induction compared to the sequence AGCGCT where induced mutagenesis was 30 to 50-fold lower.The analysis of replication events using constructions containing a strand marker across from the adduct site allowed us to demonstrate that the large difference in −2 frameshift mutagenesis is due to an intrinsic difference in the propensity of these sequences to slip during replication. How the nature of the nucleotide flanking the adduct on its 3′-side (Nb) differentially stabilises the SMI will be discussed in the light of recent structural data and theoretical models.