The need for total thyroidectomy and extended for lymphadenectomy and the need for postoperative radioiodine ablation in the treatment of papillary thyroid carcinoma is continuously debated. Since less aggressive treatment in low-risk patients has been suggested, several scoring systems have been developed to identify low-risk patients. In the current study, we compared the AMES, MACIS and TNM staging systems in predicting carcinoma-specific mortality in papillary thyroid carcinoma. Between 1967 and 1994, 495 patients with papillary thyroid carcinoma were treated at the Department of Surgery, Helsinki University Central Hospital. Carcinoma-specific mortality in the AMES low-risk group, comprising 89.7% of these patients, was 2.4%. Corresponding figures for the MACIS were 89.9%, and 2.4%, and for the TNM 85.9% and 1.2%. The mortality ratio, at 10 years, between low-risk and high-risk patients was 22.2 for the AMES, 25.0 for the MACIS and 41.8 for the TNM system. The proportion of explained variance in the Cox model was 16.3 for the AMES, 30.0 for the MACIS taken as a conitinuous variable and 28.9 for the TNM stage. The TNM stage was on average superior to the MACIS or AMES score in predicting cancer-specific mortality of patients with papillary thyroid carcinoma. This may be explained by the fact that the TNM system includes the prognostic effect of nodal metastases, which is included in neither the MACIS nor AMES systems.