During 1947-1978, 500 patients with esophageal atresia and/or tracheoesophageal fistula were treated at the Children's Hospital, University of Helsinki. The proportion of different types of anomaly followed the usual distribution, the commonest being the type with distal fistula (88.2%). For analysis, the patients were divided into five phases, each consisting of 100 consecutively treated patients. The hospital mortality decreased from 81% in the first phase to 15% in the last phase. No patients were excluded, not even those cases with lethal associated anomalies. Many more severe cases were treated in the last two phases than in the first three. Patients with distal fistula were examined separately as were the patients without a fistula and those with only a tracheoesophageal fistula. With time, early diagnosis and early referral for treatment became a rule. This led to the policy that nearly all patients in the last phase had an early operation without staging. Gastrostomy was not considered necessary when early anastomosis was possible. In the last phase, only the transpleural approach was used and single-layer end-to-end anastomosis was favored. As a whole, the type of approach or anastomosis did not have significant effect on the results, except that the Sulamaa-type end-to-side anastomosis had the highest frequency for refistula. With time there was only a slight decrease in the incidence of anastomotic leak, but significant improvement in its management. Refistula as an early complication was seen only once in the last phase. As all anastomoses were routinely dilated, severe strictures were uncommon; there were only six in the series. The factor that probably improved the prognosis the most was better pulmonary care. The improved prognosis was also due to earlier referral, modern anesthesia, and intensive care. Postoperative pulmonary complications dropped from 92% in the first phase to 40% in the last. They were the most common single cause of death in the early series but caused no deaths among the last 100 patients with no associated anomalies. Low birth weight was an important prognostic factor early in the series but in the last phase the survival rate of under-2500-gm infants with no associated anomalies was 88%. The presence of severe associated anomalies remains the most important single cause of death of an esophageal atresia patient today. Some patients (eg, trisomy 18) are beyond the possibilities of surgical treatment. To improve the prognosis of the others, efforts in the treatment of their associated (especially cardiovascular) anomalies must be made. In the treatment of esophageal atresia itself, improvement may still be achieved in the number of patients using their own esophagus without replacement procedures. Finally, a new prognostic classification of esophageal atresia patients is suggested, excluding pneumonia as an index factor in the Waterston classification.