What NICE and the NHS adviseStrong evidence
NICE's 2021 guideline offers energy management to everyone, advises no exercise programme unless it's overseen by a physiotherapist in an ME/CFS specialist team, and rules out programmes with fixed step-by-step increases such as GET. It gives its reason as "the harms reported by people with ME/CFS in the qualitative evidence, as well as the committee's experience of the effects when people exceed their energy limits". The NHS says GET is not recommended.
Sources 1, 19, 4
What the CDC advisesStrong evidence
"Standard exercise recommendations for healthy people can substantially harm patients with ME/CFS", the CDC says. It recommends pacing, staying within your "energy envelope", and keeping up the activities you can tolerate, "since deconditioning is also harmful". When people improve, activity "can be cautiously increased with monitoring for any negative effects".
Source 6
Graded exercise therapy: NICE's review of the trialsMixed evidence
NICE found 12 randomised trials of GET and rated the evidence low to very low quality. Most of the benefits were measured at around 12 weeks and weren't seen at longer follow-up, so the committee concluded there was "no clear picture of benefit". Harms were rarely measured in the trials, but people described their symptoms getting worse in studies of their experiences. Most trials didn't require PEM for entry: comparing GET with usual care, only one trial had more than 95% of people with PEM.
Source 9
The PACE trialMixed evidence
PACE (2011) included 641 adults from 6 UK clinics, chosen with criteria that didn't require PEM, and compared specialist medical care alone with medical care plus adaptive pacing, cognitive behavioural therapy (CBT) or GET. At a year, GET lowered self-rated fatigue by 3.2 points on a 0 to 33 scale, and raised self-rated physical function by 9.4 points out of 100, compared with medical care alone. People knew which treatment they had. Adaptive pacing, as tested in PACE, did no better than medical care alone. Serious reactions to treatment were rare (1% to 2%) in every group; serious adverse events were more common with GET (8% against 4%) but were judged not to be caused by it.
At a median of 2.5 years, 481 people answered follow-up questionnaires, and there was "little evidence of differences in outcomes between the randomised treatment groups". The trial team's reading is that the GET and CBT groups kept their gains while the others caught up, though 44% had had more treatment after the trial, mostly CBT or GET.
Sources 7, 20
The PACE reanalysisMixed evidence
Critics obtained the trial data under Freedom of Information and reanalysed it using the trial's original plan. On that definition, 21% improved with GET and 20% with CBT, against 10% with medical care alone, a difference that wasn't significant after allowing for the number of comparisons the plan set out. Recovery as originally defined was 4% with GET, 7% with CBT and 3% with medical care alone. Fitness on a step test, time off work and benefit claims didn't change, and the GET group walked about 30 metres further in 6 minutes than the medical care group.
The trial also changed its entry threshold for physical function partway through, from 60 or less to 65 or less, while its later "normal range" started at 60, so the two overlapped.
Sources 8, 7
The Cochrane reviewMixed evidence
Cochrane's review of 8 trials (1,518 people) found exercise therapy probably reduces fatigue at the end of treatment compared with usual care or passive treatments, by about 3.4 points on the 0 to 33 scale (moderate certainty). Whether that lasts, and whether exercise causes harm, was very uncertain. All the trials used diagnostic criteria that don't require PEM, and the authors note that people diagnosed in other ways may respond differently. Cochrane promised a full update in 2019; the project to produce it has been discontinued, so the review still rests on searches up to May 2014.
Sources 21, 22
Newer reviewsMixed evidence
A 2026 review of 17 trials (1,944 people) found exercise improved fatigue, sleep and quality of life, but not fitness or physical capacity, with very different results between trials and no difference in side effects. Its authors added a condition: "provided that individual energy limits are strictly respected to prevent harm". A 2025 review of 13 studies (1,305 people) also found less fatigue, and said the safety of exercise therapy needs further study. Both pooled older trials with newer ones, and we read only their summaries.
Sources 23, 24
What people with ME/CFS reportMixed evidence
In a survey of 1,428 people with ME/CFS, compared with other surveys totalling 16,665, between 54% and 74% said GET made them worse, while pacing was the most favoured approach, with 44% to 82% reporting benefit. In Action for ME's 2019 survey of more than 4,000 people, 88% had tried pacing in the past five years. People chose to take part, so surveys like these can't show how common harms are.
Sources 25, 26
PacingMixed evidence
NICE found 4 trials of self-management with pacing, rated moderate to very low quality, and most showed no clear difference from usual care or relaxation. In PACE, adaptive pacing was no better than medical care alone. NICE's committee recommended energy management based on its experience and what people with ME/CFS report, rather than on strong trial evidence.
Sources 9, 7, 19
Heart rate monitorsEmerging evidence
NICE lists a phone heart-rate monitor among the tools for keeping track of your activity. In an online survey of 488 people with ME/CFS, users said monitoring reduced the severity of their illness and of PEM, though they listed over 30 drawbacks as well as over 30 benefits. In a small feasibility study reported by the ME Association, 47 people tried one with no serious problems, and 89% were still using it at 8 weeks; setting the right heart rate limit was a challenge.
Sources 1, 27, 28
Very gentle exercise lying downEmerging evidence
In one Japanese trial, 48 adults whose ME/CFS hadn't improved with usual care added recumbent isometric yoga for about 12 weeks: five lying poses held very slowly with the breath, at 30% to 50% of their full strength, about 25 minutes a session, with an instructor at clinic visits and at home about 4 times a week. Fatigue and daily functioning improved more than with usual care alone, and once people were used to it there were no serious side effects, including PEM. It was one small trial, and people knew which group they were in; NICE rated an earlier isometric yoga trial low to very low quality.
Sources 29, 9
Talking therapy, medicines and other treatmentsStrong evidence
NICE says CBT may help you manage symptoms but "is not curative", should be offered only if you want it, and "does not assume people have 'abnormal' illness beliefs and behaviours" as a cause. It says not to offer any medicines or supplements to cure ME/CFS, and not to offer the Lightning Process. The NHS adds that painkillers or a low-dose antidepressant may help with pain or sleep, and that there's not enough evidence to recommend supplements.
Sources 1, 4