BACKGROUND: Lymph node status is pivotal to the staging of colorectal carcinoma. The diagnosis of a lymph node negative tumor should imply a good prognosis; however, the outcomes for Dukes' B (TNM Stage II) patients remain variable, possibly in part due to understaging. The aim of this study was to determine whether examining a specified minimum number of lymph nodes using conventional techniques would eliminate the risk of understaging and thus have an effect on prognosis. METHODS: Data on patients who underwent surgery for colorectal carcinoma at a single institution between 1985 and 1990 were reviewed. Patients with Dukes' B (TNM Stage II) or C (TNM Stage III) tumors and histologically confirmed disease-free resection margins who were treated with curative intent were included. Correlations among variables were assessed using the chi-square test, and survival comparisons were made using Kaplan-Meier curves and the log rank test. Multivariate analysis was performed using a Cox regression model. RESULTS: Dukes' B (TNM Stage II) patients with < or =6 lymph nodes examined had significantly poorer overall survival than those with > or =7 lymph nodes examined (P = 0.0014). Such a significant difference was not observed among Dukes' C (TNM Stage III) patients (P = 0.7). Survival of Dukes' C patients was significantly worse compared with that of Dukes' B patients overall and Dukes' B patients with > or =7 lymph nodes examined (P < 0.0001). There was no significant difference in survival between Dukes' C and Dukes' B patients with < or =6 lymph nodes examined (P = 0.02). The number of examined lymph nodes was the only significant parameter correlated with survival in the multivariate analysis (P = 0.002). CONCLUSIONS: Because Dukes' B patients with < or =6 examined lymph nodes have poorer outcomes than those with a higher number examined (probably due to understaging), the total number of examined lymph nodes should always be reported.