Although many attempts have been made to predict the occurrence of lymph node metastases from papillary thyroid carcinoma, there are currently no reliable means to accurately predict cervical nodal metastasis. In this study, we present a novel prediction system for the lymph node metastasis based on the histological and cyclin D1 staining features. The frequency of lymph node metastases from a series of 210 papillary thyroid carcinomas was analyzed according to the clinicopathological variables, cyclin D1 staining patterns and BRAF(V600E) mutation in tumor tissue. A total of 113 (54%) patients had lymph node metastasis. Cyclin D1 was constantly expressed at the invasive tumor front and revealed well-defined isolated glands of tumor cells in the extra-tumoral region (isolated glands) and laterally spreading tubular growth along the fibrous septa around the invasive front of the tumor (lateral tubular growth). Upon univariate analysis, an age of less than 45 years (P<0.001), tumor size of 10 mm or more (P<0.001), non-follicular variant (P=0.005), invasive growth pattern (P=0.007), extrathyroid extension (P=0.006), isolated glands (P<0.001), lateral tubular growth (P<0.001) and tumor multiplicity (P=0.005) predicted lymph node metastasis, whereas BRAF(V600E) mutation did not. Upon multivariate analysis, age (P=0.001, odds ratio (OR)=5.146), tumor size (P=0.034, OR=3.119), isolated glands (P<0.001, OR=21.042) and lateral tubular growth (P<0.001, OR=24.652) were found to be strong independent predictors of lymph node metastasis. Cyclin D1 staining of papillary thyroid carcinoma is very useful for identifying the intrathyroidal spreading or multifocality of the tumors. Tumor growth patterns verified by cyclin D1 staining can be used for the identification of papillary thyroid carcinomas with metastatic potential.