Background Most individuals with COVID-19 will recover without sequelae, but some will develop long-term multi-system impairments . The definition, duration, prevalence and symptoms associated with long COVID, however, have not been established . Methods Public Health England (PHE) initiated longitudinal surveillance of clinical and non-clinical healthcare workers for monthly assessment and blood sampling for SARS-CoV-2 antibodies in March 2020 . Eight months after enrolment, participants completed an online questionnaire including 72 symptoms in the preceding month . Symptomatic mild-to-moderate cases with confirmed COVID-19 were compared with asymptomatic, seronegative controls . Multivariable logistic regression was used to identify independent symptoms associated with long COVID . Findings All 2,147 participants were contacted and 1,671 (77.8 %) completed the questionnaire, including 140 (8.4 %) cases and 1,160 controls . At a median of 7.5 (IQR 7.1-7.8) months after infection , 20 cases (14.3 %) had ongoing (4/140 , 2.9 %) or episodic (16/140 , 11.4 %) symptoms . We identified three clusters of symptoms associated with long COVID, those affecting the sensory (ageusia, anosmia, loss of appetite and blurred vision), neurological (forgetfulness, short-term memory loss and confusion/brain fog) and cardiorespiratory (chest tightness/pain, unusual fatigue, breathlessness after minimal exertion/at rest, palpitations) systems . The sensory cluster had the highest association with being a case (aOR 5.25 , 95% CI 3.45-8.01). Dermatological, gynaecological, gastrointestinal or mental health symptoms were not significantly different between cases and controls . Interpretation Most persistent symptoms reported following mild COVID-19 were equally common in cases and controls . While all three clusters identified had a strong association with cases, the sensory cluster had the highest specificity and strength of association, and therefore, most likely to be characteristic of long COVID.