The number of episodes of diabetic ketoacidosis (DKA) is a significant outcome measure for diabetes care. We ascertained patterns of admission due to DKA over 15 yr to determine whether this indicator of diabetes care had improved in parallel with clinical practices. Between 1 January 1985 and 31 December 1999, 630 admissions were reviewed. We subanalyzed these admissions according to whether the patient was newly diagnosed, had infrequent episodes of DKA (non-relapsers) or had frequent (> or = 2/yr) episodes of DKA (relapsers). Overall there was a slight downward trend in the incidence of DKA admissions over the study period. There was a proportionate increase in the incidence of DKA amongst newly diagnosed patients, with a proportionate decrease in the incidence of DKA seen in relapsers. DKA occurring in non-relapsers remained relatively stable. Adverse clinical events during the admission were relatively uncommon and occurred in all three subgroups. There was no significant difference in HbA1C prior to admission between the relapser and non-relapser groups and there was similarity in the degree of acidosis between all three subgroups. The frequency of significant complications associated with DKA remained unchanged over the study period. Slower rehydration policies were not associated with decreases in either cerebral edema or death rates. DKA remains a significant complication of type 1 diabetes associated with a variety of significant adverse events. Our experience indicates that further efforts to reduce the occurrence of DKA must be focused upon earlier diagnosis and intervention in newly diagnosed patients.