High rates of long COVID continue to be reported in medical journals, with the prevalence in adult populations varying from 5% to 20% or higher. These high rates are further disseminated by the media and cause immense concern and anxiety in the general public. The authors of the current paper state that many scientific publications overestimate the prevalence of long COVID; they outline the reasons for these mistakes and suggest how future epidemiological research can do a better job. The authors categorize the main methodological pitfalls as follows: Overly broad case definitions. The authors of the current paper define long COVID as a syndrome or individual symptoms that are direct sequelae of being infected with SARS-CoV-2 and last for at least 12 weeks. Existing definitions from four international health organizations (including the World Health Organization in Geneva, Switzerland, and the Centers for Disease Control and Prevention in Atlanta, USA) do not require a causal link between SARS-CoV-2 infection and symptoms, meaning that any new symptoms that occur after confirmed or suspected SARS-CoV-2 infection, regardless of their aetiology, could be considered consistent with long COVID. Moreover, not all these definitions require there to be a laboratory-confirmed infection with SARS-CoV-2. Lack of controls or inappropriately-matched controls. The authors cite one recent systematic review which identified control groups in only 11% of 194 studies on long COVID. Where controls are included, they are often not properly matched to cases such as, for example, by age, sex, geography, socio-economic status and underlying health and health behaviours. Sampling bias. This type of bias occurs when certain members of a population have a higher probability of being included in a study than others. In the early stages of the pandemic, studies were often non-representative because they included fewer patients with mild symptoms or no symptoms at all. Better case definitions and appropriately matched controls will lead to better estimates of long COVID prevalence. The authors found that when they limited studies to those fulfilling these criteria, the prevalence of reported persistent symptoms in children or adults younger than 50 years of age was no different by 12 weeks post-SARS-CoV-2 infection compared with controls. Such high-quality studies stand in contrast to much of what is reported in the media and can, and should, be used to reassure the public about the risks of long COVID.