Does Paxlovid relieve long COVID symptoms?
No. In 959 adults with long COVID, neither 15 days nor 25 days of Paxlovid did better than placebo on any of the three symptom types tested.
Long COVID is the name for symptoms that hang on long after the infection has cleared. Nobody knows what causes it. One leading idea is viral persistence, the theory that pieces of the virus survive somewhere in the body and keep the immune system stirred up. If that holds, an antiviral should help. The RECOVER-VITAL trial, funded by the National Institutes of Health, tested it head on with nirmatrelvir-ritonavir, the drug sold as Paxlovid.
The trial enrolled at 69 US sites between July 2023 and September 2024. Adults qualified if they had symptoms for at least 12 weeks after a SARS-CoV-2 infection and fit one of three symptom groups: cognitive, meaning brain fog and memory trouble; autonomic, meaning dizziness on standing; or exercise, meaning crashing after exertion. Of 1,207 people screened, 959 were enrolled and analyzed, split almost evenly across the three groups. Two thirds were women and the median age was 49.
What the data show
Each person was assigned by chance to one of three arms: 15 days of the real drug followed by 10 days of placebo, 25 straight days of the real drug, or 25 days of placebo. The dose was the standard one, 300 mg of nirmatrelvir plus 100 mg of ritonavir twice a day. The question was how many people had a meaningful improvement at day 90.
Nothing moved. In the cognitive group, the 25-day course came out 3.2 percentage points ahead of placebo, a difference the trial could not tell apart from chance. The true effect very likely sits somewhere between 10.4 points worse and 16.8 points better, and there is roughly a 65 in 100 chance of seeing a gap that size by luck alone. The autonomic group came out 6.4 points worse on the 25-day course and the exercise group 7.8 points worse, both inside the range of chance. The 15-day course was just as flat, and the physical performance tests showed no differences either.
Dr. Kumar’s Take
A negative trial this large and this well built is worth more than a dozen small positive ones. The design closed the obvious escape hatches. It was placebo controlled and blinded, it ran across 69 sites, and it doubled the drug course to 25 days precisely because critics of the earlier small studies said five days was too short. Longer did not help. If anything, the 25-day arm drifted the wrong way on two of the three outcomes.
That does not close the book on viral persistence. It does say that giving this drug to people already three months or more into their illness is not the answer. Long COVID is almost certainly several different conditions wearing one name, and one drug aimed at one mechanism was always a long shot.
How strong is the evidence?
This is about as clean a test as long COVID research has produced: randomization, blinding, a placebo arm, nearly a thousand people, and a primary endpoint set in advance at day 90. The limits are still real. The group was 78 percent White, so the findings may not carry equally to everyone. Everyone had been sick for at least 12 weeks before starting, so this says nothing about treating earlier. And the main outcomes were questionnaire scores, which capture how people feel but can miss smaller biological shifts.
Safety and what it changes
No one died. There were 52 serious adverse events in 42 people, about 4 percent of the 963 who took a dose, and the investigators flagged no safety signals. Stretching the course to 25 days did not create a new problem. That matters because extended antiviral courses are sometimes handed out off label on the strength of anecdote. This trial says the longer course is tolerable, and it also says it does not help these symptoms. Tolerable and useless is still useless.
Practical takeaways
- If someone offers a long Paxlovid course as a long COVID treatment, this trial found no benefit at 15 or 25 days, so ask what evidence they are relying on.
- Symptom type did not rescue the result: brain fog, dizziness on standing, and post-exertional crashes all failed to respond.
- Care still rests on managing specific symptoms, such as pacing for post-exertional crashes and salt and fluid strategies for orthostatic symptoms.
- Joining a trial is worth considering, since this field only advances when large placebo-controlled studies get done.
Related studies and research
- Can nicotine patches treat long COVID symptoms?
- Long-term oxygen for COPD: LOTT found no benefit
- Creatine for type 2 diabetes: a placebo-controlled trial
- Single-dose psilocybin vs placebo: first double-blind depression trial
Frequently asked questions
Does this mean Paxlovid does not work at all?
No. This trial asked one narrow question: does the drug help people who already have long-standing symptoms months after infection? It says nothing about how the drug performs during an acute infection, a different problem treated in a different window. Read the result as a verdict on long COVID treatment, not on the drug itself.
Would starting Paxlovid earlier have worked better?
Possibly, and this trial cannot answer it. Everyone enrolled had been symptomatic for at least 12 weeks, so the earliest window was never tested. It is plausible that whatever drives long COVID is set in motion early and becomes self-sustaining, in which case an antiviral given months later would have nothing left to act on. That question needs its own trial.
Is viral persistence dead as an explanation for long COVID?
Not dead, but weakened. One drug at one dose for up to 25 days, in people three months or more into illness, is a specific test rather than a universal one. Virus fragments could persist in tissues this drug does not reach well, or they could be a bystander rather than the driver. What the trial does show is that the simplest version of the theory, that killing leftover virus fixes the symptoms, did not hold up.
Bottom line
RECOVER-VITAL gave 959 adults with long COVID 15 days, 25 days, or none of nirmatrelvir-ritonavir and followed them for 90 days. No symptom group improved, no secondary measure improved, and doubling the standard course changed nothing. The drug was safe, with no deaths and serious adverse events in about 4 percent of participants. Long COVID needs better ways to measure symptom burden and better treatment ideas, and a longer antiviral course is not one of them.
