The serum level of IL-6 was reported to reflect disease severity in patients with multiple myeloma. We used a specific radioimmunoassay to measure the level of IL-6 in 239 serum samples in which a monoclonal gammopathy was identified for the first time. The same sample was used for the measurement of serum C reactive protein and serum albumin. Then, an inventory of clinical and biological features allowed us to classify these patients into five groups: monoclonal gammopathy of undetermined significance (MGUS:128), multiple myeloma (MM:66), Waldenström's macroglobulinaemia (WM:27), non-Hodgkin's lymphoma (NHL:11) and chronic lymphocytic leukaemia (CLL:7). The number of patients with serum IL-6 (S-IL-6) level > 0.335 ng/ml (upper limit in normal sera) was significantly higher in the MM group (35%; Confidence Interval (CI) 23.5-46.5) compared with the MGUS group (15%; CI 8.8-21.2). The distribution of S-IL-6 levels was also significantly different between the groups (Mann-Whitney test: P < 0.01). High S-IL-6 levels were measured in 5/11 patients with NHL and 9/27 patients with WM. The distribution of S-IL-6 levels in these groups was the same as that in MGUS or MM groups. In patients with MM, elevated S-IL-6 levels were associated with haemoglobin level < 100 g/l (P < 0.005), bone marrow plasmocytosis > 50% (P < 0.005) and stages II and III in the Durie & Salmon staging system (P < 0.005). The S-IL-6 level was also related to light chain component excretion in urine (P < 0.01) and M component serum level for IgA (P < 0.01). In patients with MGUS, the S-IL-6 level correlated with serum CRP level (P < 0.05), serum lactate dehydrogenase (P < 0.05) and serum ferritin (P < 0.01). We conclude that the S-IL-6 level is a marker of high tumour burden in multiple myeloma. However, S-IL-6 level can be increased in patients with MGUS in relation to inflammatory parameters. Therefore the S-IL-6 level does not demonstrate high predictive value for the diagnosis of MM in patients with newly identified monoclonal gammopathy.