Studies have reported reduced cognitive function following COVID-19 illness, mostly from hospital settings with short follow-up times. This study recruited non-hospitalized COVID-19 patients from a general population to study prevalence of late cognitive impairment and associations with initial symptoms. We invited patients with PCR-confirmed COVID-19. A postal questionnaire addressed basic demographics, initial COVID-19 symptoms and co-morbidity about 4 months after diagnosis. About 7 months later, we conducted cognitive tests using the Cambridge Neuropsychological Test Automated Battery, comprising four tests for short-term memory, attention and executive function. We present descriptive statistics using z-scores relative to UK population norms and defined impairment as z-score <-1.5. We used multivariable logistic regression with impairment as outcome. Continuous domain scores were analysed by multiple linear regression. Of the initial 458 participants; 305 were invited, and 234 (77%) completed cognitive testing. At median 11 (range 8-13) months after PCR positivity, cognitive scores for short term memory, visuospatial processing, learning and attention were lower than norms (p≤0.001). In each domain, 4-14% were cognitively impaired; 68/232 (29%) were impaired in ≥ 1 of 4 tests. There was no association between initial symptom severity and impairment. Multivariable linear regression showed association between spatial working memory and initial symptom load (6-9 symptoms vs. 0-5, coef. 4.26, 95% CI: 0.65; 7.86). No other dimension scores were associated with symptom load. At median 11 months after out-of-hospital SARS-Cov-2 infection, minor cognitive impairment was seen with little association between COVID-19 symptom severity and outcome.