BACKGROUND Endoscopic ultrasound (EUS) is established as the most accurate method currently available for determining the depth of primary cancer invasion (T stage). Standard EUS criteria may not be accurate in assessing depth of cancer invasion and nodal status after patients have received chemotherapy or radiotherapy. METHODS We conducted a prospective study to determine whether EUS estimation of tumor size could be used to assess response to preoperative chemoradiation. Using EUS, TNM stage was assessed in 31 patients (22 men, 9 women; mean age 62 years) with cancer of esophagus or cardia (19 adenocarcinoma, 12 squamous cell cancer) before initiation of combined radiation and 5-fluorouracil/cisplatin (and/or carboplatinum) chemotherapy. The cross-sectional area of the tumor in the transverse plane at the location where the tumor had maximal thickness was calculated to estimate tumor size. EUS staging and measurement of maximal cross-sectional area were repeated at completion of chemoradiation just before surgery. Response to preoperative chemoradiation was defined as 50% reduction in maximal cross-sectional area. Surgical staging was compared between responders and nonresponders. RESULTS Eight patients who did not undergo surgery were excluded from analysis. EUST stage in the remaining 23 patients before therapy was as follows: 3 T2, 16 T3, and 4 T4. After chemoradiation, EUS T staging was changed in 6 patients (3 T4 downstaged to T3, 2 T3 downstaged to T2, and 1 T3 downstaged to T1). At surgical pathological examination, 3 patients had no residual tumor in the esophagus (T0), 5 had T1, 3 had T2, 10 had T3, and 2 had T4 tumors. EUS T staging accuracy after adjuvant therapy was only 43%. Maximal cross-sectional area decreased from a mean of 5.5 +/- 2.4 to 1.6 +/- 0.9 cm2 in responders, whereas maximal cross-sectional area went from 7.0 +/- 3.0 to 5.4 +/- 2.2 cm2 in nonresponders (p = 0.009). Ten of thirteen patients with at least a 50% reduction in maximal cross-sectional area (responders) had T0, T1, or T2 tumors at surgery, whereas 9 of 10 nonresponders had T3 or T4 tumors at surgery (p = 0.001). CONCLUSIONS (1) Standard EUS staging criteria are not accurate after neoadjuvant chemoradiation, (2) reduction in maximal cross-sectional area of tumor appears to be a more useful measure for assessing response of esophageal cancer to preoperative chemoradiation, and (3) responders have an increased likelihood of downstaging at surgery than nonresponders.