Reports of the relationship between the putative metastasis suppressor NM23 and metastasis and/or survival in colorectal cancer patients are conflicting. This study aimed to investigate whether nm23 immunostaining is correlated with established prognostic variables (Dukes' stage, degree of tumour differentiation, T stage and nodal involvement) in colorectal carcinomas for the patients treated with radical intent using Kruskal chi(2) analysis. The rates of survival at five years were estimated with the use of the Kaplan-Meier product-limit method with 95% confidence intervals derived by Greenwood's formula and the curves were compared with the use of the log-rank test. Cox proportional-hazard regression model was used to identify multivariate predictors and the corresponding outcome. The staining was performed on 112 paraffin-embedded surgical specimens collected between 1989-1992 using a monoclonal anti-nm23 antibody. Follow-up of patients was until time of death or for at least 5 years. There was not a significant correlation between tumour staging, degree of tumour differentiation, nodal involvement and nm23 status. Furthermore, there was no significant association with overall 5-year survival, disease recurrence, tumour site, age or sex. Although nm23 may be involved in suppressing tumour metastasis, nm23 immunohistochemistry has no prognostic value in colorectal cancer. For these reasons nm23 does not contribute further to the prognostic information provided by established prognostic variables.