A total of 62 canine melanocytic tumours (10 melanocytomas and 52 primary malignant melanomas) were investigated to compare the accuracy of prognosis provided by MIB-1 proliferation index (MIB-1-PI) with classical histological criteria and location. MIB-1-PI was assessed by means of quantitative image analysis of sections immunostained with MIB-1 monoclonal antibody. Tumour location, histological cell type, stromal or lymphatic vessel invasion, maximum tumour thickness, and presence of inflammation or necrosis were recorded for each case. Thirty-eight dogs were submitted to a 1-year follow-up and the clinical outcome of the disease determined. MIB-1-PI in melanocytomas differed significantly from that detected in primary malignant melanomas (P=0.0001). A significant difference in MIB-1-PI was revealed between oral and cutaneous malignant melanomas (P=0.015), and between presence and absence of lymphatic vessel invasion (P=0.05). MIB-1-PI was not correlated with the other parameters. In univariate analysis, only tumour location (oral vs cutaneous), presence of lymphatic vessel invasion, and MIB-1-PI were associated with decreased overall survival (P=0.0001,P=0.0144, and P=0.0489, respectively). In conclusion, the results of our study confirm that the assessment of the MIB-1-PI may be of additional prognostic value for dogs with primary malignant melanomas.