The kinetics of distribution of 3H methotrexate (3HMTX) in the central nervous system, plasma, and urine after intraventricular, lumbar percutaneous puncture, and spinal catheter injections were compared. Levels of 3HMTX in whole brain after lumbar percutaneous injection were 40 times less than after intraventricular injection. Injection of 3HMTX via a spinal catheter increased the level of 3HMTX in whole brain but this was still tenfold less than after direct intraventricular instillation. Also, it was found that a disproportionately high amount of 3HMTX was in the brain-stem-cerebellum region which would further reduce the concentration of methotrexate in the cerebral hemispheres. Both intraventricular and lumbar spinal catheter administration of 3HMTX produced 3HMTX levels greater than 10(-6)M (moles/kg wet weight) in spinal cord tissue as measured by 3H specific activity between 2 to 8 hours after injection. Administration by lumbar percutaneous puncture, however, rarely resulted in this suggested therapeutic level of 10(-6)M. Initial 3HMTX levels in plasma after lumbar percutaneous instillation was 24 times greater than after intraventricular or lumbar spinal catheter injections. This indicated significant and unavoidable extradural leakage after lumbar percutaneous puncture, which may account for the substantially lower levels of 3HMTX in the brain and spinal cord tissue. It is concluded that intraventricular instillation of methotrexate is the best route of administering the drug to achieve therapeutic levels of methotrexate in both whole brain and throughout the spinal cord.