An epidemic of dengue haemorrhagic fever (DHF) occurred in Delhi in 1996. A total of 240 children between the age of 4 months to 13 years of either sex, admitted in one hospital, were evaluated. Two hundred and sixteen (90%) children were from Delhi. A clinical diagnosis of dengue fever (DF) was made in 25 (10%), dengue fever with unusual bleeding (DFB) in 22 (9%), DHF in 80 (33%) and dengue shock syndrome (DSS) in 113 (47%) of the children strictly according to the WHO classification. The age peaked at 8 years. There was no association between various grades of severity of illness and age-groups though girls suffered from more severe illness. No association between severity of malnutrition and severity of illness was observed. Tourniquet test was positive in 40% with DF, 18% with DFB, 62% with DHF and 64% with DSS. In DSS haematemesis was present in 55 (49%), epistaxis in 39 (35%), melaena in 27 (24%) and ecchymosis in 34 (30%) patients. Children diagnosed as DFB had haematemesis and epistaxis in 12 (55%) and 10 (45%) respectively. Intravenous fluid requirement was clearly less in DFB patients than in DHF/DSS patients. Unusual clinical features in the form of jaundice were present in 7 (6%), hepatic encephalopathy in 6 (5%) and dengue encephalopathy in 6 (5%) patients. Dengue 2 virus was isolated from 10 of the 50 patients for whom viral culture was done on C6/36 clone of Aedes albopictus cell line. Eighteen patients suffering from DSS died giving an overall case fatality of 7.5%. The mortality rate in DHF/DSS was 9.3%. It is further suggested that DFB is a distinct entity. Most patients could be classified by the WHO classification if a retrospective packed cell volume was used to assess haemoconcentration. We suggest that development of area-specific criteria for diagnosis and management is desirable.