Coronavirus Variants

How Variants Emerge and Why It Matters

SARS-CoV-2 is a positive-sense RNA virus, and RNA viruses replicate without the proofreading machinery that DNA viruses use to catch copying errors. The result is a high mutation rate — on the order of roughly one to two mutations per genome copy per replication cycle. Most of those changes are neutral or actively deleterious to the virus. Occasionally, however, a mutation lands in a functionally important region, such as the receptor-binding domain of the spike protein, and confers a replication or transmission advantage. When that happens in an environment with sufficient host-to-host spread, natural selection does the rest: the fitter variant outcompetes existing strains and, within weeks to months, becomes dominant.

The Alpha variant (B.1.1.7), first detected in the United Kingdom in September 2020, illustrates how rapidly this process can unfold. The N501Y mutation in its spike protein increased binding affinity to the human ACE2 receptor, making it roughly 50% more transmissible than the ancestral strain — as reflected in the table above. It reached majority prevalence in the UK within about two months of first detection and spread globally before any targeted public-health countermeasures specific to it could be deployed. Delta (B.1.617.2), which emerged in India in late 2020, went further: the P681R mutation it carried, also noted in the table, contributed to roughly double the transmissibility of the original strain and increased hospitalization risk compared to Alpha, making it the most clinically severe dominant variant until Omicron displaced it.

The Omicron variant (B.1.1.529), first reported in South Africa and Botswana in November 2021, represented a qualitative shift. With 30-plus mutations in its spike protein — many clustered in the receptor-binding domain and in sites targeted by vaccine-induced antibodies — Omicron was able to reinfect individuals who had prior immunity from infection, vaccination, or both. The WHO's classification as a Variant of Concern (VOC) reflects a formal risk assessment across transmissibility, disease severity, immune evasion, and the potential to undermine diagnostics or countermeasures. As the table shows, Omicron's subvariants (BA.2, BA.5, XBB, JN.1, KP.2 and beyond) have continued to accumulate immune-evasion mutations, rotating seasonally as population immunity to preceding subvariants builds and creating new dominant strains from 2022 onward.

What changed after Omicron is not just the pace of evolution but its shape. Almost everything circulating in 2026 descends from JN.1, a BA.2.86 descendant carrying the single spike substitution S:L455S, first sampled in August 2023. Within that family, a growing share of named lineages are recombinants rather than point-mutation descendants: XFG, the backbone of much of the current US population, is a recombinant of LF.7 and LP.8.1.2, and XFZ is a recombinant of XFC.3 and PG.3.2. Recombination requires one person to be infected with two lineages at once, which is a direct consequence of high background circulation. The practical effect is that the family tree is now a mesh, and a lineage name says less about a virus's properties than it did in 2021.

CDC no longer names a single "dominant variant," and the reason is visible in its data rather than in its prose. The agency publishes estimated national variant proportions as a public dataset; below is every lineage in the most recent published period, with the 95% confidence interval CDC attaches to each estimate. Read the intervals, not just the point estimates — they are what the numbers are actually worth.

Three things that stopped

1. The Nowcast stopped. For most of the pandemic CDC published two kinds of estimate: empiric proportions based on sequences actually collected, and Nowcast projections modelling the current period before the sequencing catches up. The last Nowcast estimates in CDC's public dataset were published on April 10, 2026, covering the period ending April 11, 2026. Every publication since — May 8, June 5, July 6, July 31 and August 28, 2026 — carries empiric estimates only. In practice that means the most recent variant picture available to the public is roughly a month behind the virus rather than current.

2. The number of distinguishable lineages collapsed. Lineages below 1% of circulating virus are folded into their parent, so a shrinking table is partly a shrinking sample. The national count fell from 43 lineages (period ending July 6, 2024) to 28 (April 11, 2026) to 18 (August 1, 2026).

3. The confidence intervals widened to the point of overlapping. In CDC's file, SW.2's interval (16.4–25.8%) and XFG.1.1's (12.0–20.5%) overlap; so do the intervals of the six lineages clustered between 5% and 7%. Ranking them is not supported by the data. The honest statement is that a handful of XFG- and JN.1-descended lineages are each responsible for somewhere between a twentieth and a quarter of US infections, and no one can currently say which is in front.

This is one of the most common searches about COVID-19 variants, and it has a definite answer: no Greek letter has been assigned since Omicron in November 2021, and none of the lineages circulating today has one. Claims about a "Pi variant" or an "Omega variant" do not correspond to anything in the WHO nomenclature. How the three naming systems — WHO letters, Pango lineages and Nextstrain clades — fit together, and how to read a name like XFG or 25C, is on our variant names explained page.

The reason is procedural. WHO assigns its easy-to-say Greek labels only to variants it designates as Variants of Concern or Variants of Interest — not to every new lineage. Since March 2023 the tracking system has been organised around the reality that the landscape is dominated by Omicron descendants, and WHO's technical advisory group has not judged any subsequent lineage different enough to warrant a new top-level designation. Everything since has been tracked by its Pango lineage name (JN.1, XFG, NB.1.8.1) instead.

As of WHO's variant-tracking page, last updated 28 July 2026:

• Variants of Concern: the table is empty. WHO lists no currently circulating VOC. Alpha, Beta, Gamma, Delta and Omicron were all de-escalated as their descendants displaced them.
• Variant of Interest: JN.1 alone (designated 18 December 2023), excluding the sublineages listed separately below.
• Variants Under Monitoring (as of 27 July 2026): PQ.16.1.1 (designated 27 July 2026), NB.1.8.1 (23 May 2025), XFG (25 June 2025) and BA.3.2 (5 December 2025).

Two of the names in that list are worth noting. XFG and NB.1.8.1 acquired popular nicknames — "Stratus" and "Nimbus" — that circulated widely in news coverage during 2025 and 2026. Those nicknames are not WHO designations; they originated outside the official nomenclature and carry no risk assessment with them. BA.3.2 is the genuine oddity: the Pango network describes it as a saltation carrying roughly 40 spike amino-acid substitutions, and it descends from BA.3 rather than from JN.1 — the one lineage in current US circulation that is not part of the JN.1 family. In CDC's 2026 national estimates it has moved between about 2% and 10% of sequences without displacing the XFG family.

What the vaccine is aimed at this season

Because the leading lineages rotate, the vaccine is reformulated annually, as the influenza vaccine is. For the 2026–2027 US season, FDA's Vaccines and Related Biological Products Advisory Committee met on May 28, 2026 and recommended the JN.1-lineage XFG as the preferred variant for an updated monovalent vaccine; FDA subsequently advised manufacturers accordingly. WHO's vaccine-composition advisory group reached a closely related conclusion in its May 16, 2026 statement, recommending a monovalent LP.8.1 antigen while treating other JN.1-lineage antigens — JN.1 itself, KP.2, XFG and NB.1.8.1 — as acceptable alternatives.

The two bodies are not in conflict. XFG is a recombinant whose parents include LP.8.1.2, so both recommendations target closely related antigens within the same family. What neither can promise is a precise match: by the time doses reach pharmacies, the leading lineages will have drifted again. That is the same trade-off accepted every year for influenza, and it is the reason the formulation targets a family rather than a single strain. Our vaccines page sets out what the trial evidence behind each product does and does not establish.

What is not known

• Which lineage is actually leading in the United States right now. The most recent estimates are a month old, every one is flagged low-precision, and the top intervals overlap.
• Whether any current lineage differs in severity from the others. Detecting a severity difference requires linking sequenced infections to clinical outcomes at scale. With national estimates resting on fewer than ten sequences per lineage, that analysis is not possible from US surveillance as it now stands. Absence of evidence here is not evidence of equivalence — it is absence of measurement.
• What proportion of infections are sequenced at all. Testing is now overwhelmingly at-home antigen testing, which produces no specimen for sequencing. The sequenced sample is drawn from the shrinking minority of infections that reach a laboratory.
• Global variant distribution. WHO's tracking now rests on what member states voluntarily submit, and submission has declined with national surveillance budgets.

Compared to what?

Variant names invite a question they cannot answer on their own: how much risk does any of this represent? Two pages on this network are built to answer that rather than to track nomenclature. Our COVID-19 Risk Perspective shows each US figure beside its source and date, and leaves the cells that cannot honestly be filled — a national case count, a population-wide hospitalization rate — blank with the reason. The hub's Virus Risk Perspective compares COVID-19 with 14 other viruses on the same five measures, so the scale can be read against something.

When this site gets something wrong, the corrections log records what was published, what it says now, and the source that settled it.