<i>Background:</i> It is speculated that clinical samples do not fully reflect the characteristics of eating disorders (EDs) as they are in the general population, especially in their lowest range of severity. The present article reports the prevalence of EDs in a community sample aged >14 years, their clinical and psychopathological features, and their course and outcome on naturalistic grounds. <i>Methods:</i> The Sesto Fiorentino Study is a three-phase community-based survey where 2,355 out of 2,500 people representative of the population aged >14 years living in Sesto Fiorentino were evaluated by their own general practitioner using the Mini International Neuropsychiatric Interview plus six additional questions. All those who had positive results plus a probability sample of the non-cases were re-interviewed by psychiatrists using the Florence Psychiatric Interview. The subjects who reported ED symptoms were subsequently administered the Eating Disorder Examination (12th edition). <i>Results:</i> Overall, the lifetime prevalence of EDs was 1.21%. More precisely, 0.42% had anorexia nervosa, 0.32% bulimia nervosa, 0.32% binge eating disorder and 0.32% eating disorder not otherwise specified. All the subjects suffering from an ED fulfilled diagnostic criteria for at least another DSM-IV axis I psychiatric disorder. At the moment of the interview, conducted a few years (average 7 years) after the onset of the disorder, 50% had fully recovered from EDs, 26.9% were currently affected by an ED, 23.1% showed a persistent body image disturbance and/or the presence of compensatory behaviours. <i>Conclusions:</i> Community surveys conducted by clinicians may provide useful additional information on the psychopathological features, natural course and outcome of these disorders on naturalistic grounds.