How Vaccines Work

• Immunity has two layers: a fast, generic innate response and a slow, specific adaptive one. Vaccines train the second.
• The lasting product of a vaccine is not antibodies — it is memory B and T cells.
• Five vaccine families, from weakened live pathogens to genetic instructions the body reads itself.
• Boosters exist for three distinct reasons, and "the vaccine wore off" is only one of them.
• The herd-immunity threshold is approximately 1 − 1/R₀ — which is why measles needs ~95% and most diseases need far less.

The two immune systems

Your body defends itself in two quite different ways, and confusing them is the source of most misunderstanding here.

The innate immune system is the standing garrison. Skin and mucous membranes, plus cells like macrophages and neutrophils that recognise broad molecular patterns common to whole classes of microbe. It responds within minutes to hours, it is the same response every time, and it does not learn. Fever, inflammation and the general awfulness of the first day of an illness are largely innate.

The adaptive immune system is the specialist unit. Its cells recognise a specific molecular shape — an antigen — and it works by selection: your body maintains an enormous randomly generated library of B and T cells, each with a receptor for a different shape, and an infection finds the few that happen to match and orders them to multiply. That process, clonal expansion, is why the adaptive response is powerful but slow. It takes roughly one to two weeks to get going from a standing start.

Within it, three players matter here:

• B cells mature into plasma cells that mass-produce antibodies — proteins that bind the pathogen, neutralising it or tagging it for destruction.
• Helper T cells (CD4+) coordinate: they license B cells to produce better antibodies and direct the rest of the response.
• Cytotoxic T cells (CD8+) kill the body's own cells once they are infected, which antibodies cannot do because antibodies work outside cells.

The decisive part is what happens afterwards. Once an infection clears, most of the expanded cells die off, but a subset persists as memory cells. On a second encounter, those memory cells are already present, already specific, and already numerous. The response arrives in days rather than weeks, and it is larger. Frequently the pathogen is cleared before you notice symptoms.

A vaccine is a way of generating that memory without the first illness. That is the entire idea.

The five families

Different pathogens need different tricks, which is why "vaccine" describes a category of tools rather than a single technology.

Type	How it works	Examples
Live attenuated	A weakened strain that replicates but does not cause disease	MMR, varicella, yellow fever, oral polio
Inactivated	The whole pathogen, killed	Inactivated polio (IPV), hepatitis A, rabies
Subunit / recombinant	Isolated antigens, often a single protein	Hepatitis B, HPV, acellular pertussis, shingles
Toxoid	An inactivated bacterial toxin, not the bacterium	Tetanus, diphtheria
Nucleic acid / viral vector	Genetic instructions; your cells make the antigen	mRNA COVID-19 vaccines, rVSV Ebola vaccine

The trade-off across this table is consistent: the closer a vaccine is to a real infection, the stronger and longer-lasting the immunity, and the more caution it requires. Live attenuated vaccines are the closest thing to the real event and tend to produce robust, long-lived immunity — often lifelong from one or two doses. They are also the ones generally not given to people who are significantly immunocompromised or pregnant, precisely because they replicate.

Subunit vaccines sit at the other end. Showing the immune system one purified protein is very safe and very controllable, but a lone protein with no signs of danger around it is easy to ignore. That is what adjuvants are for: substances such as aluminium salts, or newer formulations like MF59 and AS01, that create enough local inflammatory signal for the immune system to take the antigen seriously. An adjuvant is essentially a way of saying this matters to a system that would otherwise shrug.

The toxoid family solves a different problem. Tetanus does not make you ill by growing everywhere — it makes you ill by producing an extraordinarily potent neurotoxin. Immunity to the bacterium is beside the point; you need antibodies against the toxin. So the vaccine is the toxin, chemically inactivated so it can be recognised but cannot act.

What mRNA vaccines actually do

The newest family is worth spelling out mechanically, because it is the most misunderstood and the biology is not complicated.

The vaccine contains messenger RNA — the same class of molecule your cells use constantly to carry instructions from DNA to the protein-building machinery — packaged in a lipid nanoparticle, a tiny fat bubble that protects the fragile mRNA and helps it get inside a cell. Once inside, ribosomes in the cytoplasm read it and build the encoded protein, in this case a viral surface protein. The cell displays that protein, the immune system sees an unfamiliar antigen accompanied by signs of disturbance, and mounts the adaptive response described above.

Then the mRNA is degraded by ordinary cellular enzymes over a period of days. It is a transient instruction, not an installed component.

Three points follow directly from where this happens:

1. The mRNA never enters the nucleus. DNA is in the nucleus; translation happens in the cytoplasm. They are different compartments.
2. Human cells cannot write RNA back into DNA. That requires reverse transcriptase, an enzyme our cells do not deploy for this purpose, plus integration machinery this mRNA does not carry.
3. Nothing persists. The protein is displayed and cleared; the mRNA degrades. What remains afterwards is memory cells — exactly what remains after any other vaccine.

The technology looked sudden in 2020 but was not. Its critical enabling discovery came in 2005, when Katalin Karikó and Drew Weissman showed that chemically modifying the nucleosides in synthetic mRNA stopped it from provoking a crippling innate inflammatory reaction — the problem that had blocked the field for years. They were awarded the Nobel Prize in Physiology or Medicine in 2023 for that work.

Why boosters exist

Three separate reasons hide behind one word, and distinguishing them clears up most of the confusion.

Antibodies wane; memory lasts longer. Circulating antibody levels fall over months and years — this is normal and not a failure. Memory B and T cells persist far longer, sometimes for life. So declining antibody titres can mean reduced protection against any infection while protection against severe disease is largely retained. It is why a vaccine can stop preventing a mild case while still preventing a hospitalisation, which looks like a contradiction only if you think antibodies are the whole story.

Some schedules are multi-dose by design. The hepatitis B and HPV series are not boosting a fading response; the first dose primes and later doses drive the response to full strength. Each exposure also improves antibody quality through affinity maturation, not just quantity.

Some pathogens move. Influenza viruses accumulate mutations in their surface proteins fast enough that last year's antibodies bind poorly — antigenic drift. The annual flu shot is therefore reformulation, not boosting: a new composition selected each year based on global surveillance. This is also why measles, whose virus is antigenically stable, gives durable protection from two childhood doses while influenza never will.

Where the herd-immunity number comes from

Community immunity is the indirect protection that arises when enough people are immune that a pathogen cannot find a chain of susceptible hosts. It is what protects newborns too young to be vaccinated, people undergoing chemotherapy, and the small fraction in whom a vaccine simply did not take.

The threshold is not a policy choice. It follows from how contagious the disease is, captured by R₀ — the average number of people one infectious person would infect in a fully susceptible population. The approximation is:

threshold ≈ 1 − 1/R₀

Disease	Approximate R₀	Approximate threshold
Measles	12–18	~92–95%
Pertussis	12–17	~92–94%
Polio	5–7	~80–86%
Smallpox	5–7	~80–85%
Seasonal influenza	1–2	~30–50%

This single line of arithmetic explains why measles is always the first disease to return when coverage slips. At an R₀ near 15 the threshold sits above 90%, so a community can lose only a few percentage points of coverage before transmission becomes self-sustaining again. Diseases with lower R₀ tolerate far more slack.

Real populations are messier than the formula: immunity is unevenly distributed, people mix in clusters rather than at random, and a national average of 95% conceals neighbourhoods at 70% where outbreaks actually start. The formula gives the right order of magnitude and the right intuition, not a guarantee.

How we know they work

The evidence base has an unusual property: the effect is often large enough to be visible without statistics. Diseases that filled hospital wards within living memory became rare within a decade of a vaccine's introduction, in country after country, with the drop tracking the year of introduction rather than the year of improved sanitation or antibiotics.

The most direct demonstration is smallpox. A disease that killed an estimated 300 million people in the twentieth century alone was declared eradicated by the WHO in 1980, through a targeted vaccination campaign. It no longer exists outside two secure laboratories. Wild polio has been pushed to a handful of cases in a small number of countries by the same approach.

At scale, a 2024 analysis published in The Lancet estimated that fifty years of the WHO's Expanded Programme on Immunization — running since 1974 — has averted on the order of 154 million deaths, the majority of them in children under five.

Safety monitoring does not stop at licensing. Vaccines pass through phased clinical trials before approval, and then into permanent post-marketing surveillance systems — VAERS and the Vaccine Safety Datalink in the United States, the Yellow Card scheme in the UK, EudraVigilance in the EU — designed to detect rare adverse events that trials of tens of thousands of people are too small to see. That system has real teeth: it is how the rare clotting syndrome associated with certain adenovirus-vector COVID-19 vaccines was identified within months, leading several countries to restrict their use.

It is worth stating plainly that the claimed MMR–autism link originated in a 1998 Lancet paper that was retracted in 2010, and whose lead author was struck off the UK medical register for serious professional misconduct. Subsequent studies have looked hard and found nothing, including a Danish cohort following 657,461 children published in 2019.

Vaccines are not risk-free — nothing effective is. What the evidence supports is that for the diseases on routine schedules, the risks are small, well-characterised, actively monitored, and vastly outweighed by the risks of the diseases themselves.

Further reading: The WHO maintains accessible explainers on how vaccines work and vaccine types. The 154-million estimate is Shattock et al., The Lancet (2024). For the immunology in depth, Janeway's Immunobiology is the standard text and is freely readable via NCBI Bookshelf.