As a scientist who works every day on the immunology of Covid-19 and long Covid, I’m well aware that, heading into autumn and the return to school, the UK faces yet more Covid confusion and disharmony. Where are we headed next? Isn’t it over? And why keep harping on about mitigation when we now have so many other concerns? Any discussion of our current Covid situation must consider the legacy of disability and misery associated with long Covid. In my opinion, there is now some good news among the old bad news. Over the past few months, Office for National Statistics data shows the estimated number of people with long Covid beginning to fall, from a peak of 2 million in May to about 1.8 million. I take this to mean that some are gradually recovering. And while long Covid following Omicron BA.5 infection is clearly happening, new cases of long Covid are appearing at a lower frequency. Colleagues in Singapore, a country with a large peak of Omicron infections following a relatively mild early pandemic, mention talk of quiet long Covid clinics without patients. There is also some indication we may be getting closer to more precisely defining and treating long Covid. Many studies around the world have been set up to recruit groups with long Covid to compare them with “rapid recovery” cases – people who recovered quickly and fully from Covid – to try to find differences in levels of antibodies, hormones, immune cells or other things that can be measured with a blood test. These so-called “defining biomarkers” can be gamechangers. They can help health services define and refer cases, provide more extensive evidence for employers and tribunals, and also point towards identification of therapies and treatments. One of the first such studies was reported this month in a preprint from Akiko Iwasaki, David Putrino and colleagues at Yale. They report a clear biomarker delineating differences in the long Covid group, with signals including low serum cortisol (a hormone involved in control of the stress response) and evidence of reactivation of latent Epstein-Barr virus. This is not yet an outright diagnostic test for long Covid, but it expands our knowledge of what exactly is happening behind the symptoms, as well as signposting potential treatments. Despite some glimmers of good news, it bears repeating that long Covid continues to be a source of largely unaddressed despair, especially for those “first-wave” two-year-plus long-haulers, unable to return to work and in many cases, getting short-shrift from employers. The cruel irony of this is self-evident for a patient group in which our much-applauded “frontline heroes” are massively overrepresented. Long Covid remains a very real risk, but the best way to avoid it is to avoid getting infected (or reinfected) in the first place. The initially successful UK vaccine rollout of 2020-21 is ancient history in terms of the current battle. The BA.5 subvariant has so many immune-evasive mutations that it is a distant relative of the ancestral strain of Sars-CoV-2 against which those first vaccines were generated. In any case, most people’s neutralising antibody levels have waned pretty much to baseline – or similar to unvaccinated – levels, even if they are triple vaccinated. Thus the huge burden of breakthrough infections and reinfections. There is strong consensus around the pressing need for autumn boosters. Immunologists and vaccine producers have been engaged in considerable debate about the complexities of ensuring that the next generation, variant-specific boosters are the best they can be. At the international level, approaches have diverged. The UK has procured doses of a bivalent – meaning targeted at two strains – booster, carrying the original, ancestral Sars-CoV-2 sequence as well as the BA.1 sequence, the Omicron variant that was with us at the end of 2021. In the US, meanwhile, the government has ordered 170m doses of a bivalent vaccine targeting the original, ancestral Sars-CoV-2 sequence, and the currently relevant BA.5 sequence. Data from trials of these approaches is still thin. The current sense is that boosting against BA.1 may not offer strong cross-protection against BA.5. But data for BA.5 specific boosters is also minimal. How much do such details matter? A team in Sydney, led by Miles Davenport, analysed data from eight other studies to model the effectiveness of variant-specific boosters. The data there is reassuring. Even the original first generation booster is predicted to increase protection from symptomatic infection in a six-month period from 50% to 85.6%. The new, variant-modified boosters, generally an additional 1.5 times more potent in terms of Omicron neutralisation, would provide over 90% protection. There will probably be some differences between boosters once more studies are completed. But this complexity should never erode public confidence that you (and the clinically vulnerable around you) will be safer this autumn and winter with a booster, any booster, than without. There should be no equivocation about a full-on booster programme. Indeed, we should remember that UK vaccine uptake, at 80%, sits behind other European countries such as France, Italy and Portugal, with lower vaccine uptake skewed to lower socioeconomic groups in whom Covid-19 is often more severe. We need to increase vaccine uptake in the unvaccinated, along with a strong message from the government and the Joint Committee on Vaccination and Immunisation that protection and normalisation now depends on widespread and efficient rollout of the new boosters. Those following the data and also mindful of protecting the clinically extremely vulnerable will also be wearing masks and maximising ventilation. Danny Altmann is professor of immunology at Imperial College London, a trustee of the Medical Research Foundation and of Long Covid Support, and co-author of The Long Covid Handbook