Immune imprinting or original antigenic sin (OAS) originally referred to a phenomenon of suboptimal immune response to a repeat exposure to a virus that was antigenically distinct from the original virus infection. OAS has been implicated in higher mortality in young people during the 2009-10 H1N1 pandemic where the elderly (H1N1 exposure in childhood) appeared relatively well protected compared to younger individuals whose first influenza infection was not H1N1. Immune imprinting is part of a rapid recall system and is highly effective against a slowly evolving virus (drifting) but not antigenically shifting viruses such as influenza and SARS CoV-2. As predicted by OAS, suboptimal neutralization responses to the highly divergent SARS-COV-2 lineage Omicron have been observed in animal models and individuals previously vaccinated with primary course of ancestral (Wu-1) vaccine. Due to the rapid scale up of vaccine before emergence of the antigenically distinct Omicron variant, it is unknown whether immunological imprinting for occurs in the context of SARS-COV-2 infection itself. We longitudinally assessed humoral responses to primary two dose Ad26.COV2.S Wu-hu-1 based vaccination in a Nigerian population following the global emergence of Omicron. At study entry in Jan 2023, we found 93% and 58% of pre-vaccination participants previously exposed to ancestral Wu-1 and Omicron virus respectively by anti-N IgG and anti-receptor binding domain (RBD) IgG Wu-1 and Omicron -specific antibodies. In participants with no evidence of prior exposure to Omicron, neutralisation against Wu-1 was significantly higher than Omicron variants as expected. However, serum neutralisation titres in participants who were anti-RBD Omicron IgG positive were paradoxically 2-fold lower for Omicron BA.1 as compared to Wu-1. This is clear evidence for imprinted immunity from the ancestral pre-omicron lineage viruses, and remarkably these old responses to Wu-1 were able to dominate over more recent, likely multiple, Omicron lineage infections. Furthermore, in these participants with prior exposure to Omicron and evidence of imprinting, we observed that further Omicron infection and Wu-1 based vaccine was associated with boosting of responses across variants with equalisation of neutralisation titres for Wu-1 and Omicron variants. However, omicron responses did not surpass ancestral responses, suggesting persistence of imprinting and only partial mitigation. Although neutralization responses at high titres were observed post dose 1 vaccination against ancestral and Omicron variants BA.1, BA.2, BA.4 in nearly all participants, neutralisation against the highly immune evasive XBB recombinant variant remained substantially lower, with a second vaccine dose providing very modest boosting. These data highlight immune imprinting against SARS-CoV-2 prior to vaccination and its persistence thereafter. In present day unvaccinated populations where serum neutralisation responses to pre-Omicron variants dominate, use of an omicron variant based vaccine should be used in preference to Wu-1 based vaccine to override imprinting and provide broader protection for vulnerable populations such as the elderly or those with compromised immunity.