Intrathecal production of kappa free light chains occurs in multiple sclerosis and can be measured using the kappa free light chain index. Kappa free light chain index values can be determined more easily than oligoclonal bands detection and seem more sensitive than the immunoglobulin (Ig)G index to diagnose multiple sclerosis. We assessed the value of oligoclonal bands, kappa free light chain index cut-offs 5.9, 6.6 and 10.61, and IgG index to diagnose multiple sclerosis with prospectively acquired data from a clinically isolated syndrome inception cohort. We selected patients with sufficient data to determine oligoclonal bands positivity, MRI dissemination in space and time, IgG index and sufficient quantities of paired CSF and blood samples to determine kappa free light chain indexes (n = 214). We used Kendall's Tau coefficient to estimate concordance, calculated the number of additional diagnoses when adding each positive index to dissemination in space and positive oligoclonal bands, performed survival analyses for oligoclonal bands and each index with the outcomes second attack and 2017 MRI dissemination in space and time and estimated the diagnostic properties of oligoclonal bands and the different indexes for the previously mentioned outcomes at 5 years. Oligoclonal bands were positive in 138 patients (64.5%), kappa free light chain-5.9 in 136 (63.6%), kappa free light chain-6.6 in 135 (63.1%), kappa free light chain-10.61 in 126 (58.9%) and IgG index in 101 (47.2%). The highest concordance was between oligoclonal bands and kappa free light chain-6.6 (τ = 0.727) followed by oligoclonal bands and kappa free light chain-5.9 (τ = 0.716). Combining dissemination in space plus oligoclonal bands or kappa free light chain-5.9 increased the number of diagnosed patients by 11 (5.1%), with kappa free light chain-6.6 by 10 (4.7%), with kappa free light chain-10.61 by 9 (4.2%) and with IgG index by 3 (1.4%). Patients with positive oligoclonal bands or indexes reached second attack and MRI dissemination in space and time faster than patients with negative results (P < 0.0001 except IgG index in second attack: P = 0.016). In multivariable Cox models [adjusted hazard ratio (95% confidence interval)], the risk for second attack was very similar between kappa free light chain-5.9 [2.0 (0.9-4.3), P = 0.068] and kappa free light chain-6.6 [2.1 (1.1-4.2), P = 0.035]. The highest risk for MRI dissemination in space and time was demonstrated with kappa free light chain-5.9 [4.9 (2.5-9.6), P < 0.0001], followed by kappa free light chain-6.6 [3.4 (1.9-6.3), P < 0.0001]. Kappa free light chains-5.9 and -6.6 had a slightly higher diagnostic accuracy than oligoclonal bands for second attack (70.5, 71.1 and 67.8) and MRI dissemination in space and time (85.7, 85.1 and 81.0). Kappa free light chain indexes 5.9 and 6.6 performed slightly better than oligoclonal bands to assess multiple sclerosis risk and in terms of diagnostic accuracy. Given the concordance between oligoclonal bands and these indexes, we suggest using dissemination in space plus positive oligoclonal bands or positive kappa free light chain index as a modified criterion to diagnose multiple sclerosis.