Tumour cells in malignant melanomas express molecules associated with tumour progression; however, up until now, no marker has been able to identify the tumour cells from which metastases are derived. It has recently been shown that in human melanoma cell lines, populations expressing peanut agglutinin (PNA)-binding glycoproteins are able to generate metastases, and that such cells do exist in primary human melanomas, their presence being associated with the degree of local invasion that governs the metastasis risk. To further investigate the correlation between the expression of PNA-binding glycoconjugates by cells from primary melanomas and the patient's individual risk of recurrence or metastasis, a molecular epidemiological approach employing histochemical techniques within a case-control design was developed. The main objective of this study is to determine whether an histochemical staining with the lectin PNA of cells in the primary lesion is associated with an increased risk of local recurrence of metastasis, and with survival, independently of Breslow's tumour thickness. The study comprises the comparison of the PNA labelling index and of the type and intratumour location of the staining as a function of clinical outcome in two matched series of patients with known clinical outcome: patients who had died at 5 years and patients alive at 5 years (to assess association with survival), and patients who experienced a recurrence within the first 5 years and patients alive without recurrence at 5 years (to assess association with risk of recurrence). A matched case-control design was used with a variable number of controls matched to each case. Apart from age (+/-5 years), sex and centre where diagnosis was made, matching was made on histogenetic type and primary tumour thickness (four categories being considered: <0.75, 0.76-1.5, 1.51-3 and >3 mm).