Imagine two people sharing a room with someone who has influenza. One becomes ill; the other does not. Does that prove the virus caused the illness, or that the decisive factor was something inside the host?
It proves neither on its own. This familiar observation contains two different questions: what initiates an infection? and what determines its outcome? The first asks about a causative agent; the second asks about immunity, exposure, health and environment. Confusing them makes both germ theory and its critics look simpler than the evidence allows.
The short answer: viruses are identifiable infectious agents, not merely cellular waste. But infection and disease are not synonyms. The host and the circumstances of exposure help determine whether a viral encounter produces no apparent illness, a mild one, or a severe one.
Why the alternative sounded reasonable
An earlier version of this article entertained a radical version of the “terrain” hypothesis: perhaps viruses were cell-derived debris, or extracellular vesicles such as exosomes, and apparent outbreaks reflected a shared environmental insult rather than transmission of an infectious agent. That claim has been removed because it does not survive the strongest experimental tests.
There was a legitimate observation behind the question. Most bacteria living with us are not enemies; many are part of normal microbial communities. The same pathogen may produce very different outcomes in different people. And electron microscopy can reveal particles that look surprisingly similar, even when they have different biological roles.
The leap from those observations to “viruses do not exist” is the error. Resemblance is not identity, and variation in outcome does not demonstrate absence of a cause.
What would actually distinguish a virus from cellular debris?
A picture of a small round particle is not sufficient. Neither is one positive molecular test. A stronger identification combines independent methods: the particle’s molecular components, its reproducible behaviour in susceptible cells, the sequence of its genome and, where appropriate, its ability to initiate a new infection.
Here the question becomes experimentally sharp. If a proposed virus is only inert cell debris, then supplying a defined viral genetic system to suitable cells should not reproducibly generate new infectious particles carrying the specified genetic changes. Yet reverse-genetics studies have done just that.
Experiment 1: constructing infectious virus from genetic instructions
In 1999, researchers reported generating influenza A viruses from cloned DNA components in cultured cells. They could also introduce chosen genetic changes and recover viruses carrying them. This was not a photograph interpreted as a virus: it was a controlled intervention linking specified viral genetic material to the appearance of infectious progeny. [1]
A 2020 study similarly described assembling and recovering SARS-CoV-2 from an infectious complementary-DNA clone, enabling the genome to be modified and its effects examined. These experiments depend on host-cell machinery, as viruses do, but their results cannot be reduced to an undifferentiated release of cellular waste. [2]
What this establishes: there are specific, manipulable viral genetic systems capable of generating infectious particles. What it does not establish: that every detected fragment is infectious, or that every infection inevitably causes symptoms.
Experiment 2: giving a measured viral exposure to volunteers
Human challenge studies add a different kind of evidence. Researchers deliberately expose carefully selected volunteers to a specified virus under medical supervision and then measure infection, symptoms, immune responses and sometimes recoverable viable virus. These studies are ethically restricted and their findings cannot automatically be generalised to older or medically vulnerable populations.
In a SARS-CoV-2 challenge study published in Nature Medicine in 2022, 36 unvaccinated young adult volunteers were inoculated. After two were excluded from the per-protocol analysis because of evidence of infection before inoculation, 18 of 34 met the study’s infection criteria: 53% (95% confidence interval, 35–70%). Viable virus could be recovered from nasal samples, and symptoms were mostly mild to moderate. [3]
A 2026 influenza A/H3N2 aerosol-challenge pilot studied 14 volunteers. In the higher-dose groups, infection occurred in 3 of 4 participants using one delivery system and 2 of 4 with another. Those small denominators matter: the study demonstrates experimental infection under its conditions, not a universal infection probability of 50–75%. [4]
A larger historical synthesis of influenza challenge experiments found that infection was common among volunteers with low pre-existing haemagglutination-inhibition antibody titres, while symptoms and shedding varied. Across studies, doses, strains and outcome definitions differed, so the pooled estimate should not be treated as a prediction for an individual. [5]
What these studies establish: a defined viral exposure can initiate measurable infection, and infectious virus can subsequently be recovered. What they also demonstrate: the same exposure does not produce an identical outcome in everyone.
But exosomes and viruses really do overlap
Extracellular vesicles (EVs) are membrane-bound particles released by cells. They carry proteins, lipids and sometimes nucleic acids. “Exosome” names a particular biogenesis route, but many experiments cannot establish that origin with confidence, which is why EV researchers often use broader terminology.
Some enveloped viruses and EVs overlap in size, density and membrane composition. They can use related cellular pathways; infected cells may release a mixture of conventional vesicles, virions and hybrid particles. This is a real technical problem for purification and identification, not a reason to discard virology. The MISEV2023 methodological guidance explicitly warns about separation and characterisation limitations. [6]
A 2024 study on single-particle protein and RNA analysis, including HIV, illustrates the distinction. Virions and non-viral vesicles can be physically similar, and some vesicles contain viral components. Researchers therefore examined viral genomic RNA and other features rather than trying to identify the particle by shape alone. The relevant distinction is a functionally infectious viral system, not a perfectly unique size or silhouette. [7]
An important caution follows: viral RNA inside a vesicle is not, by itself, proof that the vesicle is an infectious virion. Nor does an ordinary exosome become a virus simply because it carries a fragment of viral material. Contamination, mixed particle preparations and non-replicating hybrids must be considered before assigning function.
What a PCR test can and cannot tell us
PCR and related assays detect selected nucleic-acid sequences. A positive result can be strong evidence that target genetic material is present, assuming the assay and sample are reliable. But detection alone does not measure how sick the patient is, guarantee that viable infectious virus remains, or determine how readily that person could transmit it at that moment.
That limitation is sometimes used to make an invalid second claim: because PCR can detect non-infectious remnants, viruses themselves must be imaginary. The conclusion does not follow. Viral identification rests on converging evidence, including sequence-specific interventions, culture, challenge experiments, particle characterisation and epidemiological patterns. PCR is one instrument in that collection, not the entire argument.
Where ‘terrain’ belongs in the model
The sharpest correction to a simplistic “one germ, one inevitable disease” picture comes from ordinary infectious-disease biology itself. The epidemiological triad distinguishes agent, host and environment. Infection depends on the agent and a susceptible host meeting under enabling conditions; clinical illness adds another layer of host response and tissue effects. [8]
- Agent: viral strain, inoculum, route of entry and biological properties.
- Host: pre-existing immunity, age, genetics, physiological state and underlying disease.
- Environment: ventilation, duration of exposure, crowding, humidity and other relevant circumstances.
These components interact rather than compete. Even an identical viral inoculum may lead to different results because the volunteers are not biologically identical. Conversely, a robust immune response does not make an infectious agent fictitious.
The microbiome offers an especially instructive bacterial example. In a randomised trial of recurrent Clostridioides difficile infection, donor faecal microbiota transplantation produced clinical resolution in 20 of 22 participants (90.9%) versus 15 of 24 (62.5%) given their own stool. That study shows that altering microbial ecology can alter disease outcomes. It does not show that C. difficile is unreal, and faecal transplantation is a regulated clinical procedure rather than a do-it-yourself treatment. [9]
Which claims survive the evidence?
- “Viruses are fictional.” Contradicted by genome-directed recovery of infectious particles and controlled infection experiments.
- “Some viral particles resemble exosomes.” Supported, but morphology and particle size are insufficient for classification.
- “PCR positivity always means contagiousness.” Unsupported. Infectiousness requires additional context and, in research, sometimes viability measurements.
- “The host and environment shape disease.” Supported. This complements, rather than replaces, infectious causation.
- “Because the host matters, pathogen-specific treatment or prevention is unnecessary.” Does not follow from the evidence.
What remains open
There are important unanswered questions about how extracellular vesicles influence infection, which host factors best predict severe illness, and how to distinguish infectious virus from residual genetic material in particular clinical settings. These are testable questions. They do not require pretending that viral genomes, cultured infectious particles or controlled transmission have never been demonstrated.
A useful challenge to any explanation is to ask for its risky prediction. What observation would make the hypothesis fail? A model that calls every new finding “cell debris” regardless of genome, infectivity and experimental manipulation has little discriminatory power. A model that predicts different outcomes after a specific genetic change or a controlled exposure can be tested and potentially rejected.
The conclusion
Viruses are real, but the formula “virus present = person ill” is wrong. The strongest model is not a contest between germs and terrain. It is a causal system in which identifiable infectious agents interact with host biology and exposure conditions. The scientific improvement is to measure each part instead of using one to erase the others.
Editorial note (11 October 2026): An earlier version of this article proposed that viruses might be cellular debris or exosomes. This revision retains that hypothesis as a question to test, explains why the evidence rejects its strong form, and separates direct observations from interpretation. No independent medical review is claimed for this revision.
Sources and methods
The evidence above prioritises controlled experiments and primary methods papers. Volunteer challenge studies have restricted populations and are not general-purpose estimates of individual risk; vesicle studies have recognised isolation and classification limitations.
- Neumann G, et al. (1999). Generation of influenza A viruses entirely from cloned cDNAs. PNAS.
- Xie X, et al. (2020). An Infectious cDNA Clone of SARS-CoV-2. Cell Host & Microbe.
- Killingley B, et al. (2022). Safety, tolerability and viral kinetics during SARS-CoV-2 human challenge in young adults. Nature Medicine.
- Rouphael N, et al. (2026). Feasibility and Safety of Aerosolized Influenza Virus Challenge in Humans Using Two Modern Delivery Systems. Journal of Infectious Diseases.
- Carrat F, et al. (2008). Time lines of infection and disease in human influenza: a review of volunteer challenge studies. American Journal of Epidemiology.
- Welsh JA, et al. (2024). MISEV2023: Minimal information for studies of extracellular vesicles. Journal of Extracellular Vesicles.
- Simultaneous Protein and RNA Analysis in Single Extracellular Vesicles, Including Viruses. (2024). ACS Nano.
- CDC: Principles of Epidemiology, epidemiological triad.
- Kelly CR, et al. (2016). Effect of Fecal Microbiota Transplantation on Recurrence in Multiply Recurrent C. difficile Infection. Annals of Internal Medicine.