A further improved chromatographic method for the simultaneous determination of the total amount of ODAP, selectively the amount of its neurotoxic form, beta-ODAP, and free L-glutamate in raw Lathyrus sativus (grass pea) seed samples is described using post-column refractive index in combination with bioelectrochemical detection. The biosensor is based on crosslinking horseradish peroxidase (HRP) and an Os-containing mediating polymer with poly(ethyleneglycol)(400) diglycidyl ether (PEGDGE), forming an inner hydrogel layer and then immobilising L-glutamate oxidase (GlOx) as an outer layer on top of a graphite electrode. Addition of polyethylenimine (PEI) to the hydrogel is believed to have sensitivity and stability enhancing effect on the biosensor. The double-layer approach in the biosensor construction avoided direct electrical wiring of GlOx and resulted in a higher sensitivity of 4.6 mA/M cm2 with respect to beta-ODAP and a wider linear range (1-250 microM) for both L-glutamate and beta-ODAP when compared with a single-layer approach where GlOx, HRP, and Os-polymer are crosslinked together. The limit of detection for the chromatographic-biosensor system was found to be 2 microM with respect to beta-ODAP and 0.7 microM with respect to L-glutamate. The refractive index detection on-line with the biosensor enabled full control of the chromatographic system for the determination of the total amount of ODAP, selectively the amount of beta-ODAP and L-glutamate. Ten grass pea samples have been collected from Lathyrism prone areas of Ethiopia to test the applicability of the presently developed analytical system for real sample analysis. The toxin levels of grass pea collections were determined in an aqueous extracts and ranged from 0.52 to 0.76%, dry mass basis. Comparison of results of an established spectrophotometric assay and that of the present system has shown an extraordinary degree of agreement as revealed by parallel "t" test (90% confidence limit). The present system has operational stability of more than 50 h. Analysis time per sample is 10 min after extraction for 90 min.