Information collected in a clinical study on a random sample of 99 patients with inoperable lung cancer, treated with radiotherapy, was compared to the staging information in the Maastricht cancer registry. Validity of sex (0% disagreements), date of birth (0%), histology (1% major disagreements) and treatment (1%) was high, but the validity of stage was lower: 12% major and 23% minor disagreements. The misclassification of stage did not result in a shift in the survival estimates. If cancer registries intend to use stage in comparisons of survival, more validation studies are necessary.