Since 1992, various combinations of thoracoscopy (VATS), laparoscopy or hand-assisted thoracolaparoscopy have been used for 'minimally invasive' cancer esophagectomy (MIE). Despite widespread current use, indications and potential benefits of the many technical approaches remain controversial. A systematic literature search was conducted until June 2007. Out of 128 publications, 46 original series (1932 patients) met the inclusion criteria and were analyzed for surgical and oncological outcome. No prospective controlled study has compared any MIE technique to another or to open surgery. Most publications are retrospective series of highly selected patients, mostly excluding high-risk patients and locally advanced (T3) tumors. Altogether, the overall conversion rate was 5.9%, mortality 2.9% and morbidity 46%, many papers reporting only major complications. Overall, rates for pulmonary complications were 22%, leakage 8.8% and vocal cord palsy 7.1%. Fifteen tracheo-bronchial injuries or fistulas (1% of all VATS cases) were reported. Laparoscopy and VATS were combined in 11 series (609 patients, 4.7% conversions, 2.4% mortality). VATS combined with (mini)-laparotomy was reported in 14 papers (743 patients, 6.3% conversions, 2.4% mortality). Laparoscopy combined with right thoracotomy was reported in four papers (147 patients, 5.4% conversions, 2% mortality). Laparoscopic transhiatal resections were reported in 17 papers (433 patients, 7% conversions, 4.6% mortality). Overall morbidity rates for these four approaches were 43%, 47.6%, 51.6% and 46%, respectively. Data on oncological outcome are scarce. Lymph node retrieval (median of all series: 14 nodes, range 5-62) was mostly inferior to open surgery standards and follow-up too short to draw definitive conclusions regarding long-term survival. Based on the available literature, the morbidity and mortality of MIE is substantial and not inferior to radical open esophagectomy in experienced centers. Many different operative techniques for MIE have been reported without obvious superiority for any of them. The term 'minimally invasive' is not supported by hitherto reported results. Selection bias and huge variability in extent of resection and lymphadenectomy impair comparisons of different MIE techniques. Oncological outcome of MIE remains largely unknown by lack of good quality data and selection bias. MIE remains an investigational and still evolving treatment for invasive cancer.