Objective: We used a population-based cancer registry to examine the association between lymph node counts and mortality to determine the minimum number of lymph nodes that should be examined as part of esophageal resection. Methods: Using the Surveillance Epidemiology and End Results database, we identified patients who had an esophagectomy for invasive esophageal carcinoma from 1988 through 2005 and who had a known number of lymph nodes examined pathologically. After stratifying patients (0, 1-11, 12-29, and 30 or more lymph nodes examined) based on a recursive partitioning analysis, we assessed the association between lymph nodes counts and mortality using the Kaplan-Meier method. To adjust for potential confounding covariates, we used a Cox proportional hazards regression model. Results: Of the patients in the Surveillance Epidemiology and End Results database with esophageal cancer, 4882 met our inclusion criteria. We noted a significant difference between the lymph node groups with regards to unadjusted all-cause (P < .0001) and cancer-specific mortality (P = .004). After adjusting for cancer registry, patient factors, tumor characteristics, and timing of radiation therapy, we noted a significant difference between the lymph node groups with regards to all-cause and cancer-specific mortality. Compared with patients who had no lymph node evaluation, only patients who had more than 12 lymph nodes examined had a significant improvement in mortality; patients who had 30 or more lymph nodes examined had significantly lower mortality rates than the other groups. Conclusion: To maximize all-cause and cancer-specific survival, esophageal cancer patients should have at least 30 lymph nodes examined pathologically as part of esophageal resection.