Herpes zoster is rare in otherwise healthy children, but it is more common in association with immunosuppression. Maternal varicella infection during pregnancy and varicella occurring in the newborn represent risk factors for childhood herpes zoster. However, some controversies persist about risk factors, diagnosis, and the natural history of childhood disease. In a 2-year prospective study, 18 children with herpes zoster were clinically diagnosed in outpatient consultations in a hospital dermatology unit. Data about age, dermatome involvement, underlying disease, and history of previous varicella were recorded. Tzanck smears, biopsy specimens, and sera were obtained from 18, 4, and 10 children, respectively. The varicella zoster virus major envelope glycoprotein gE was detected in 16 of 18 smears and all four biopsies. Herpes simplex virus I was demonstrated in one of the smears. The established risk factors for childhood herpes zoster were only found in one child. Evidence for previous full-blown varicella and varicella with few lesions was recorded in 7 and 4 of the 17 immunocompetent children, respectively. No history of varicella was recalled in 6 of 17 cases, although a serologic clue of past varicella infection (IgM negative, IgG positive) was disclosed. Recurrent herpes zoster was diagnosed in one immunocompromised child. Zoster-associated pain was localized and the disease severity remained mild in all children. Established risk factors for childhood herpes zoster were only rarely found in our series of patients. In contrast, unrecognized varicella and varicella with few lesions were frequently recorded and may represent additional risk factors for shingles in childhood. Zosteriform herpes simplex virus infections should be differentiated from childhood herpes zoster, emphasizing the importance of precise viral identification.