When a tumor invades the muscularis mucosa and submucosal layer (T1a-MM and T1b in Japan), esophageal squamous cell cancer poses 10-50% risk of lymph node metastasis. By this stage of esophageal cancer, surgery, although very invasive, is the standard radical therapy for the patients. Endoscopic mucosal resection (EMR) is the absolutely curable treatment for cancer in the superficial mucosal layer. Because of its minimal invasiveness, the indications of EMR may be expanded to include the treatment of T1a-MM and T1b esophageal carcinoma. To date, the clinical outcomes of EMR for T1a-MM and T1b patients have not been fully elucidated. Here, the retrospective analysis of the clinical outcomes is reported. Between January 1994 and December 2007, 247 patients underwent EMR at Kanagawa Cancer Center. Of these individuals, 44 patients with 44 lesions fulfilled the following criteria: (i) extended EMR treatment for clinical T1a-MM and T1b tumor; (ii) diagnosis of clinical N0M0; and (iii) follow up for at least 1 year, and negative vertical margin. These patients were reviewed for their clinical features and outcomes. Statistical analyses were performed by the Kaplan-Meier methods, the Chi-square test, and the Cox proportional hazard model. P-value of <0.05 was considered statistically significant. The data were analyzed in February 2009. Based on the informed consent and their general health conditions, 44 patients decided the following treatments immediately after the EMR: 2 underwent surgery, 1 underwent adjuvant chemotherapy, and 41 selected follow up without any additional therapy. Of the 41 patients, 20 selected this course by choice, 12 because of severe concurrent diseases, 2 because of poor performance status, and 7 because of other multiple primary cancers. Twelve patients died; two were cause specific (4.5%), eight from multiple primary cancers, one from severe concurrent diseases, and one from unknown causes. No critical complications were noted. Median follow-up time was 51 months (12-126). Five patients ultimately developed lymph node metastasis. One patient with adjuvant chemotherapy required surgery, and another was treated with chemotherapy whose subsequent death was cause specific. The other three patients received chemoradiotherapy and have not shown cause-specific death. Overall and cause-specific survival rates at 5 years were 67.3% and 91.8%, respectively. Among 41 patients treated by EMR alone, only one died from primary esophageal cancer (2.4%), and overall and cause-specific survival rates at 5 years were 75.6% and 97.6%, respectively. Multivariate analysis revealed that severe concurrent diseases including multiple primary cancers and the administration of 5-fluorouracil-based chemotherapy for multiple primary cancers significantly influenced survival (P= 0.025, hazard ratio [HR] 13.1 [95% confidence interval 1.5-114]) and (P= 0.037, HR 0.213 [95% confidence interval 0.05-0.914]), respectively. Eight and six patients developed metachronous esophageal squamous cell cancer and local recurrence, respectively. With the exception of one patient, they could be retreated endoscopically. EMR is a reasonable option for the patients with T1a-MM and T1b esophageal carcinoma without clinical metastasis, especially for the individuals with severe concurrent diseases. The prognostic factors for the benefit of EMR in such cases should be further examined.