The aims of this study were to evaluate the prognostic predictabilities of the risk classifications (Waterston, Montreal, and Spitz), preoperative treatment of pneumonia and the length of esophageal gap in esophageal atresia. Twenty-five cases of esophageal atresia were reviewed retrospectively. The data was analyzed with Likelihood ratio test for trend and Fisher's exact test. Survival rates were 100.0% in group A and B, and 60.0% in group C according to Waterston classification (p=0.027). By Montreal classification, 95.7% in group I and 50.0% in group II (p=0.086). By Spitz classification, 95.5% in group I and 66.7% in group II (p=0.159). The survival rate was 85.7% in the patients with pneumonia preoperatively treated and 94.4% in the patients without pneumonia. The difference was not significant (p=0.490). Survival rate in the patients with long gap (> or = 3cm) was 50.0% and 100.0% in patients with short gap (< 3cm) with significant difference (p=0.020). The Waterston classification could be modified into two groups because the prognosis of both group A and B were the same. Preoperative pneumonia does not seem to be a risk factor. The length of the esophageal gap showed prognostic value.