Specimens from 1007 patients with suspected neurological disturbances had quantitative and qualitative measurements made of cerebrospinal fluid and serum to investigate the presence of locally synthesised IgG. Qualitative measurement was recorded as the presence or absence of oligoclonal banding, and the quantitative measurement was derived by the use of the IgG index, the log index and the Reiber, Schuller and Tourtellotte formulae. The patients were divided into two categories, on the basis of banding: those with local synthesis and those without. The sensitivity, specificity and efficiency for each of the quantitative measurements were then calculated. Receiver-operator curves were also constructed for each of the quantitative measurements. 282 samples showed local synthesis of IgG by isoelectric focusing, whereas the best quantitative assay (log index) could only detect 198. Therefore, we conclude that oligoclonal banding should be adopted as the standard laboratory measurement of local synthesis of IgG in the diagnosis of neurological disorders, and that the diagnostic use of quantitative measurements should be abandoned for routine purposes. Furthermore, we suggest that quantitative analysis, at its current level, is misleading and has little value in the understanding of neurological disorders, but may be of use in serially monitoring individual patients as part of their therapeutic trials.