The presenting clinico-hematologic features of 386 patients with nonmyelomatous monoclonal gammopathy (MG) were correlated with the frequency of malignant transformation to evaluate the most important variables conditioning its evolution into multiple myeloma (MM) or Waldenstrom macroglobulinemia (WM). Most of the patients (335) had monoclonal gammopathy of undetermined significance (MGUS: 39 IgA, 242 IgG, 54 IgM): the remaining 51 patients (12 IgA, 39 IgG) fulfilled all of the MGUS diagnostic criteria (according to Durie) except that bone marrow plasma cell (BMPC) content was 10% to 30%, and so they were defined as having monoclonal gammopathy of borderline significance (MGBS). There were no significant differences between the MGUS and MGBS groups in terms of age, sex, or median follow-up. After a median follow- up of 70 and 53 months, respectively, 23 of 335 MGUS and 19 of 51 MGBS patients had undergone a malignant evolution. Univariate analysis of the IgA and IgG patients showed that the cumulative probability of the disease evolving into MM correlated with diagnostic definition (MGBS v MGUS), BMPC content (> or = 10% v < 5% and < or = 5% v > 5%) and reduced serum polyclonal Ig. In the IgG cases, there was also a significant correlation with detectable Bence Jones proteinuria, serum monoclonal component (MC) levels and age at diagnosis (> 70 v < = or 55 years). In the IgG cases as a whole, the same variables remained in the Cox model where the BMPC percentage was considered after natural logarithmic transformation and the monoclonal component as g/dL value. The relative risks of developing MM are the following: 2.4 for each 1 g/dL increase of IgG, serum MC, 3.5 for detectable light chain proteinuria, 4.4 for the increase of 1 unit in log. BMPC percentage, 6.1 for age > 70, 3.6 and 13.1 for a reduction in one or two polyclonal Ig. In conclusion, our study allows the identification of a particular subset of MGUS patients (MC < = or 1.5 g/dL, BMPC < 5%, no reduction in polyclonal Ig and no detectable light chain proteinuria) at very low- risk of evolution, who can be considered as having benign monoclonal gammopathies. We also describe a previously undefined group of MG patients (with monoclonal gammopathy of borderline significance) who are at high-risk of malignant evolution. These findings could have a considerable impact on the cost/benefit ratio of monitoring programs in these patients.