The predictive power of serum kappa/lambda ratios on initial presentation of immunoglobulin G (IgG) or IgA monoclonal component was studied to differentiate between monoclonal gammopathy of undetermined significance (MGUS) and multiple myeloma (MM) patients. The retrospective study involved 145 patients clinically diagnosed with monoclonal gammopathy of undetermined significance or multiple myeloma, who had serum M-protein IgG <35 g/L or IgA <20 g/L at M-protein detection. Serum light chains kappa and lambda were measured by fixed-time nephelometry. Test performance indices, predictive values and likelihood ratios were calculated according to the Weissler recommendation. MM patients were considered as diseased and MGUS patients as non-diseased in order to estimate the performance characteristics of serum kappa/lambda ratios. There was a statistically significant difference in kappa/lambda ratios distribution between both groups of patients, in both M-protein kappa-type (Mann-Whitney U=168, p<0.001) and in M-protein lambda-type (Mann-Whitney U=143, p<0.001). Negative likelihood ratios at threshold levels of 0.6 and 4.2 were 2.17- and 3.32-fold greater, respectively, than positive likelihood ratios, so that the predictive power of a serum kappa/lambda ratio within these limits is better in ruling out (negative predictive power) than ruling in disease (positive predictive power). The post-test characteristics of a serum kappa/lambda ratio interval between 0.6 and 4.2 in discriminating MGUS from MM in our geographic population were: sensitivity 0.96 (0.93-0.99 95% CI); specificity 0.70 (0.63-0.77); positive predictive value 0.68 (0.64-0.73); negative predictive value 0.96 (0.94-0.99); likelihood ratios (+)LR 3.23 (2.68-4.04); and (-)LR 17.16 (11.00-63.00). Thus, serum M-protein with a kappa/lambda ratio between 0.6 and 4.2 increases the posterior probability of MGUS from 0.60 to 0.96 in asymptomatic patients, for whom only monitoring may be suggested when the serum kappa/lambda ratio is within these limits.