The present study aimed at investigating whether in gastric cancer patients stage migration occurs with extension of lymphadenectomy, when node metastases are staged according to the new pN classification (UICC 1997). The investigation involved 921 patients, who underwent R0 gastric resection for gastric cancer between 1988 and 1998 in three different Italian centres: Verona (n=236), Forlı`(n=409), Siena (n=276). The relation among lymphadenectomy and pN category was assessed by Kendall's partial rank-order correlation coefficient, controlling for depth of tumour invasion. A direct evaluation of the Will Rogers phenomenon was accomplished in the Verona series, by comparing the number of positive nodes actually observed with the number of positive nodes which would have been retrieved by a less extended lymphadenectomy (D1). The number of positive nodes increased remarkably with the enlargement of lymphadenectomy, especially in pT2 patients (from 2.2+3.9 in D1 to 3.9+5.0 in D3) and in pT3/pT4 patients (from 5.1+5.9 in D1 to 11.3+12.6 in D3). Non-parametric statistics highlighted a weak (Kendall's partial T=0.128) but significant (P50.001) correlation between pN category and extension of lymphadenectomy. In the direct analysis of the Verona series, 22 patients out of 230 (9.6%) migrated to a lower pN tier when ignoring positive nodes retrieved from the second and third level. This percentage increased to 39.1% (90 out of 230) when adopting the TNM 87 classification. In conclusion stage migration is of minor importance in gastric cancer patients, staged according to the new pN classification. In the past, all staging systems for gastric cancer defined N stage by location of node metastases relative to the primary (Hermanek and Sobin, 1992; JRSGC, 1995). In order to simplify the pN classification, to reduce methodological errors and to achieve a higher reproducibility, the UICC has presented in 1997 a new classification system based on the number of involved nodes (Sobin and Wittekind, 1997).In the last 3 years many authors compared the new classification with the previous one and showed that both the site and the number of the positive lymph nodes are independent prognostic factors in gastric cancer patients; however the pN categories (especially pN2 and pN3) based on the number are prognostically more homogeneous. Moreover the new TNM system seems to identify a group of high risk patients (pN3 category) with poor long term survival Roder et al, 1998;de Manzoni et al, 1999;Yoo et al, 1999;Katai et al, 2000;Klein Kranenbarg et al, 2001).Some authors maintain that the number of positive lymph nodes does not increase further when more than 15 lymph nodes are excised (Hermanek, 1991;Siewert et al, 1993). If this statement was correct, another advantage of the new pN classification would be the prevention of stage migration (the so-called Will Rogers phenomenon) (Feinstein et al, 1985) which affected the previous pN classification, as shown by Bunt et al (1995).To verify the independence of the new pN classification fro...