This time in 2020, we watched with horror as the realities of the pandemic and its death toll unfurled. Most hardly dared imagine that effective vaccines might appear in a fraction of the time taken for previous efforts, effectively stemming the pandemic tide. But despite the success of the vaccines in greatly reducing the odds of hospitalisation or death, viral evolution had plenty more to throw at us. The onslaught of highly immune-evasive variants was, for most of us in immunology and virology, unforeseen. We’d come to think of the coronavirus family as being rather more stable – less error-prone in terms of mutations – than many viruses. And we had never before had to roll out relatively new ways of developing vaccines, involving mRNA or recombinant adenoviruses, at this scale and in the heat of battle. Having started out brilliantly, the real-life state of play today is self-evidently suboptimal. The vaccines rapidly induce hugely high levels of protective, neutralising antibodies in most people, but these levels wane within months of each sequential dose. Meanwhile, Omicron and the subvariant BA.2 have managed to mutate almost every amino acid residue targeted by protective antibodies, escaping protection. And so you have the unhappy equilibrium currently endured by the UK: more than 300,000 new cases a day, as of late last week, and a continuing caseload of more than 3 million, with hospital admissions and excess deaths holding steady at a new – high – setpoint. All this despite one of the highest vaccination rates in the world. We are living in a precarious truce imposed through frequent mRNA boosters to keep the viral caseload “manageable”. But there are signs this isn’t sustainable, and that a strategy simply consisting of boosters in perpetuity may not be fit for purpose. Recent case surges in Hong Kong, Denmark and Scotland emphasise the fragility of that balance. And new evidence from the past two years suggests that encounters with different variants of Covid or different vaccine types can alter the effectiveness of later jabs in surprising ways – an effect called immune imprinting. This raises the possibility that booster performance could be even less predictable and effective in the future. Sars-CoV-2 began as a single variant, which we term the Wuhan strain. But we now inhabit a world where no two people share precisely the same exposure history: we have never been infected, or were asymptomatically, mildly or severely infected during any or a combination of the Wuhan to Alpha, Delta, Omicron or BA.2 waves, and we’ve all had somewhere from zero to four doses of diverse vaccines. The combination of these exposures gives each of us a unique immune memory repertoire. Imagine a huge jar of pills of different colours, each especially good for responding to a given present or future variant. Someone whose experience has been an Alpha infection plus three doses of Pfizer may have brilliantly built up lots of green pills at the expense of others. But this is less good for you if the next variant mainly needs yellow pills. It turns out the order and type of exposure can affect how our immune system responds later on. In a recent paper reported in the journal Science, we compared protective immunity between people infected in the first wave with the original strain and in the second wave with the Alpha variant. In second wave-infected people, encounters with an Alpha infection plus two vaccine doses gave lower protective (known as neutralising) antibody responses against the Wuhan and Beta variant, yet higher responses against Delta. Given the number of vaccines and strains, these interactions are unpredictable, but will shape how our immunity holds for future waves. It needs more investigation. These are complex problems demanding careful research, long-term planning, trials and even some intelligent crystal ball-gazing. We must evaluate many approaches. Some places have announced a fourth dose rollout for first generation Pfizer vaccines (which cross-neutralises recent variants, but very suboptimally); some vaccine makers have pivoted to targeting the Omicron spike; others are working on polyvalent vaccines to include several different versions of spike, or clever structural approaches to target those parts of spike that would be the same across all past and future variants, and maybe even across those coronaviruses still awaiting crossover from bats and pangolins. This latter approach is exciting and the subject of recent efforts across many teams, including research trials through the US National Institutes of Health and at Cambridge University. There are also advanced programmes considering intranasal – nose – vaccination to achieve local mucosal immunity, increasing the chances of blocking transmission at that site altogether, and vaccine platforms that could be much more durable. The take-home message is that the pandemic is very much with us and evolving dynamically, with a long, bumpy road ahead. The option to sleepwalk through this, taking automatic-pilot choices based on what was “good enough” in the first wave is one we adopt at our peril. We must look at options besides simply boosting through every successive wave. At a time when the US has cut future vaccine research funding, and the UK also needs to maintain its momentum, this should be an urgent priority. Danny Altmann is a professor of immunology at Imperial College London, who has contributed advice to the Cabinet Office, APPG on long Covid, and the EU