Risk factors for postacute sequelae of SARS-CoV-2 infection ("long COVID") in community-dwelling populations remain poorly understood. Large-scale data, follow-up, comparison groups, and a consensus definition of long COVID are often lacking. Using data from the OptumLabs Data Warehouse on a nationwide sample of commercial and Medicare Advantage enrollees from the period January 2019 through March 2022, we examined demographic and clinical factors associated with long COVID, using two definitions of people who suffer symptoms long after they were first diagnosed with COVID-19 ("long haulers"). We identified 8,329 long haulers using the narrow definition (diagnosis code), 207,537 long haulers using the broad definition (symptom based), and 600,161 non-long haulers (comparison group). On average, long haulers were older and more likely female, with more comorbidities. Among narrow-definition long haulers, the leading risk factors for long COVID included hypertension, chronic lung disease, obesity, diabetes, and depression. Their time between initial COVID-19 diagnosis and diagnosis of long COVID averaged 250 days, with racial and ethnic differences. Broad-definition long haulers exhibited similar risk factors. Distinguishing long COVID from the progression of underlying conditions can be challenging, but further study may advance the evidence base related to the identification, causes, and consequences of long COVID.