The response of the central nervous system to boron neutron capture irradiation, with either p-boronophenylalanine (BPA) or borocaptate sodium (BSH) as neutron capture agents, has been assessed using a rat spinal cord model. The mean latency times for the development of myelopathy after irradiation with the thermal neutron beam-alone, or in combination with BPA or BSH, were 23.7 +/- 0.3, 21.8 +/- 0.4 and 19.6 +/- 0.4 weeks, respectively. The radiation-induced lesion in the spinal cord was characterised by white matter necrosis. Due to the variations in the microdistribution of different neutron capture agents in body tissues, it was considered inappropriate to define the biological effectiveness of the high LET radiation, resulting from the 10B(n, alpha)7Li neutron capture reaction, relative to photon radiation, using the term 'relative biological effectiveness' (RBE). The term 'compound biological effectiveness' (CBE) factor was used as an alternative. ED50 values for the various irradiation modalities were calculated from probit fitted dose effect curves. Expressed as total physical absorbed doses these values were 13.6 +/- 0.4, 30.3 +/- 2.7 and 13.8 +/- 0.5 Gy after irradiation with the thermal neutron beam alone, or the thermal neutron beam in combination with BSH or BPA, respectively. The RBE of the thermal neutron beam was 1.4 +/- 0.04. The microdistribution of the two neutron capture agents played a crucial role in the determination of the overall biological effect, after thermal neutron activation. BSH, which is excluded from the CNS parenchyma by the blood brain barrier, had a low CBE factor value of 0.46 +/- 0.5. BPA, on the other hand, which crosses the blood brain barrier and distributes in the CNS parenchyma, had a higher CBE factor value of 1.33 +/- 0.16.