Paxlovid in 2026: What the Updated Efficacy Data Actually Shows

The Evolving Paxlovid Picture

When Paxlovid received emergency authorization in late 2021, it was a turning point. The original EPIC-HR trial showed an 89% relative risk reduction in hospitalization and death among unvaccinated, high-risk adults. Those numbers made headlines for good reason.

Four years later, the clinical picture looks different. Most American adults have been vaccinated, boosted, or previously infected — often all three. The dominant variants have shifted multiple times. And we now have real-world data from hundreds of thousands of patients, not just the 2,246 in the original trial.

The question I hear most often in 2026 is no longer "does Paxlovid work?" It's "does it work for me?" That answer depends on your age, your vaccination status, your underlying health, and what you're trying to prevent. Here's what the current evidence actually says.

What the Real-World Data Shows

The single largest study to date used the N3C (National COVID Cohort Collaborative) database to evaluate Paxlovid outcomes in 703,647 patients across 34 US medical centers.^[1] Published in PLoS Medicine in January 2025, the results showed a 39% reduction in hospitalization (95% CI 36–41%, P<0.001) and a 61% reduction in death (95% CI 55–67%, P<0.001).

Those numbers held up in both vaccinated and unvaccinated patients. The drug still works against current Omicron subvariants. But here's what the top-line numbers don't capture: the absolute benefit depends heavily on baseline risk.

For a 70-year-old with diabetes and heart disease, a 39% reduction in an already elevated hospitalization risk translates to a meaningful number of avoided hospital stays. For a healthy 35-year-old who is fully vaccinated, the baseline risk is already so low that a 39% relative reduction amounts to a very small absolute difference.

A Science magazine analysis of the available trial data put this in sharper focus.^[2] In clinical trials that enrolled mostly vaccinated adults (median age in the 40s), Paxlovid's effect was not statistically significant. The best-case estimate was a 1.3% absolute improvement — compared to 5.5% in the original unvaccinated trial. Vaccinations may have already done enough that Paxlovid can't add much more for standard-risk patients.

In my practice, this means the conversation has shifted. I'm not asking "should you take Paxlovid?" as a yes-or-no question. I'm asking: what is your individual risk profile, and does the expected benefit outweigh the inconvenience, cost, and drug interactions?

The Rebound Question: Settled Science

Viral rebound after Paxlovid made major news in 2022 and 2023. Patients would feel better after five days of treatment, then test positive again — sometimes with returning symptoms. The worry was understandable: did the drug just delay the infection?

We now have solid data on this. A Harvard/Mass General study of 142 participants found that 20.8% of Paxlovid recipients experienced virologic rebound, compared to just 1.8% of untreated patients.^[3] So rebound is real, and it's clearly more common with treatment. Those patients also shed live, potentially contagious virus during the rebound window, with prolonged viral shedding lasting around 14 days versus fewer than 5 days in the non-rebound group.

But here's the part that matters most: rebound episodes were mild. They did not lead to hospitalization or severe disease.

The next logical question — "should I take another course of Paxlovid if I rebound?" — was answered by a phase 2 randomized controlled trial published in Clinical Infectious Diseases in October 2025.^[4] In 436 participants, a second 5-day Paxlovid course after rebound did produce faster viral RNA decline, but no clear clinical benefit. Zero COVID-related hospitalizations or deaths occurred in either the retreatment or placebo group. Retreatment was well tolerated, but it was unnecessary.

What I tell patients: if you experience rebound, you likely do NOT need another course. Isolate per current guidelines, manage symptoms, and expect it to resolve on its own. For immunocompromised patients, the picture may be different — extended courses of 10 to 15 days have shown reduced rebound rates in this specific group.^[5]

Resistance: Not Yet a Concern

With millions of Paxlovid courses prescribed worldwide, resistance was always a theoretical worry. Antiviral drugs can select for resistant viral strains, and COVID mutates rapidly. So where do we stand?

A September 2024 study in JAMA Network Open analyzed samples from 156 participants and found that nirmatrelvir resistance mutations were rare, transient, and low-frequency.^[6] The mutations were not linked to virologic rebound and didn't persist in a way that would suggest accumulating resistance over time.

The researchers also analyzed GISAID surveillance data — the global database that tracks SARS-CoV-2 sequences — and found no increase in resistance-associated mutations in the US population over time. This is reassuring. It suggests that current prescribing patterns are not driving meaningful resistance.

One unexpected finding stood out: a mutation that confers resistance to ensitrelvir (a different protease inhibitor used in Japan) actually increases susceptibility to nirmatrelvir. The two drugs have distinct enough binding profiles that cross-resistance isn't a given. For now, resistance remains a theoretical concern, not a clinical one.

Long COVID Prevention: The Disappointing Data

Early in the pandemic, there was hope that treating acute COVID aggressively might prevent Long COVID. The logic seemed sound: reduce viral replication early, reduce the downstream inflammatory cascade that drives persistent symptoms.

The data hasn't supported this. A large RECOVER/N3C cohort study published in PLOS Medicine in September 2025 used causal inference methodology to test whether Paxlovid reduced subsequent Long COVID incidence.^[7] The result: no significant effect in the overall population.

There were small signals in subgroups. Patients 65 and older saw a 12% reduced risk. Those aged 18–49 had a 7% reduction. But the numbers needed to treat were high, and the effects fell short of clinical significance for most patient groups.

This is worth being direct about: Paxlovid is not a Long COVID prevention tool. If you're taking it, the reason should be reducing your risk of hospitalization and death during acute infection — not protecting against symptoms that might or might not develop weeks later.

Evidence at a Glance

What This Means for You

The evidence supports a targeted approach. Paxlovid still offers clear benefit for specific groups, but it's no longer a blanket recommendation for everyone who tests positive.

A few practical points. Paxlovid must be started within 5 days of symptom onset — the sooner, the better. This is one area where telehealth genuinely helps: a virtual visit can get you evaluated and prescribed within hours of testing positive, without waiting for an in-person appointment.

Drug interactions with ritonavir (the booster component in Paxlovid) are real and potentially serious. Ritonavir inhibits CYP3A4, which means it affects the metabolism of dozens of common medications — certain blood thinners, statins, immunosuppressants, sedatives, and more. Tell your doctor about every medication you take. This is not a "take two and call me in the morning" situation.

Cost is also a factor. Without insurance, a 5-day course runs $1,400 to $1,800.^[8] Most commercial insurance plans cover it with a $0–$75 copay. Medicare Part D covers it. Pfizer's PAXCESS co-pay card and patient assistance programs can help if cost is a barrier.

Last reviewed: Apr 2026 by Parth Bhavsar, MD. This article is for informational purposes and does not constitute medical advice. Consult your physician for treatment decisions.