are vaccines safe?

The short version: no blanket answer exists. Some vaccines clearly help some people; every vaccine injures a few. The system is built to count the first and miss the second.

The question "are vaccines safe" has no single answer because vaccines are not one thing. A flu shot in a healthy adult is not the same as an mRNA shot in a teenage boy, which is not the same as a rotavirus dose in an infant. The evidence here shows the split clearly. On the benefit side, a Norwegian study of 4.6 million adults found that people with three or more mRNA doses had about 60% lower all-cause death rates than unvaccinated people during 2021–2023 [2]. A Hungarian study of 6.4 million adults found all six COVID vaccines reduced death after adjusting for the healthy-vaccinee effect [8]. A Danish study of 101,786 heart-failure patients found no increased risk of worsening heart failure, myocarditis, or blood clots within 90 days of mRNA vaccination [7]. Those are real, large, and worth taking seriously.

On the harm side, the same evidence base shows the injuries. A BMJ living review found myocarditis after mRNA vaccines occurs in about 50 to 139 per million doses in boys aged 12–17, mostly after the second dose, mostly in males, and mostly within a few days [17]. A Korean study found 21 deaths among 480 myocarditis cases after mRNA vaccination [21]. A US county-level study of over 3,000 counties found that higher vaccine uptake tracked with higher all-cause excess mortality in 2022 and 2023, even after adjusting for prior deaths [5]. A UK analysis of official statistics found that vaccinated people's death rates relative to unvaccinated people rose steadily over time, crossing the unvaccinated line between September 2022 and January 2025 depending on age [3]. These studies are ecological or observational, so they cannot prove causation. But they are the opposite of reassuring.

The funding question matters here. The Norwegian and Hungarian studies are independent government analyses, which is why their benefit findings carry weight. The VAERS-based safety studies are passive surveillance, which detects almost nothing by design; underreporting is the norm, and a safety claim built on VAERS is built on the absence of reports in a system designed not to receive them [38]. The most informative design, vaccinated versus unvaccinated over years with hard clinical endpoints, was run in Norway and Hungary and showed benefit. The same design run in the UK and US showed a different picture emerging over time. Both cannot be fully right, and the difference is exactly where the uncertainty lives.

My call: for older adults and high-risk people, the evidence for net benefit is strong. For young males, the myocarditis risk is real and the benefit is smaller. For everyone, the long-term safety data does not exist, because the studies that would show it have not been run long enough. Confidence: moderate.