Eight histopathologists, based at different hospitals, who had previously examined 100 consecutive colposcopic cervical biopsies were circulated with the results of the initial study. The slides were then 'reblinded' and re-examined by the pathologists who, as before, assigned them into one of six diagnostic categories. The degree of interpathologist agreement for the seven observers who returned usable responses was characterized by kappa statistics and compared to the corresponding figures for the same observers from the previous study. Although some of the observers showed significant alterations in their diagnostic practices there was persistent poor agreement for CIN 1 and 2, mediocre agreement for CIN 3 and excellent agreement for invasive carcinoma. Intra-observer agreement was consistently better than inter-observer agreement for each of the diagnostic categories. Significant differences were found among observers in the degree of intra-observer variability. The 20 cases in which there was most disagreement were re-examined by one of the authors who compared these with 20 biopsies which caused little disagreement. Disagreement was considered to be associated with florid papilloma-virus changes, basal cell hyperplasia and severe inflammation in varying combinations. On the basis of these findings we suggest changes in the terminology of CIN lesions.