Management decisions for patients with cancer are frequently taken within the context of a multi-disciplinary team (MDT). There is little known, however, about decision-making at team meetings and whether MDT decisions are all implemented. This study evaluated team decision-making in upper gastrointestinal cancer. Consecutive MDT treatment decisions were recorded for patients with oesophageal, gastric, pancreatic and peri-ampullary tumours. Implementation of MDT decisions was investigated by examining hospital records. Where decisions were implemented it was recorded as concordant and discordant if the decision changed. Reasons for changes in MDT decisions were identified. 273 decisions were studied and 41 (15.1%) were discordant (not implemented), (95% confidence interval 11.1\20.0%). Looking at the reasons for discordance, 18 (43.9%) were due to co-morbid health issues, 14 (34.2%) related to patient choice and 8 (19.5%) decisions changed when more clinical information was available. For one discordant decision, the reason was not apparent. Discordant decisions were more frequent for patients with pancreatic or gastric carcinoma as compared to oesophageal cancer (P = 0.001). Results show that monitoring concordance between MDT decisions and final treatment implementation is useful to inform team decision-making. For upper gastrointestinal cancer, MDTs require more information about co morbid disease and patient choice to truly optimize the implementation of multi-disciplinary expertise.