A study was performed to determine whether an implanted, inductively coupled nuclear magnetic resonance (NMR) imaging spine coil could provide a significant gain in signal-to-noise ratio (SNR) on images of the spinal cord relative to the SNR of optimized surface coils. Implanted coils were surgically affixed to the upper lumbar spine (first lumbar through third lumbar vertebrae) in a total of four adult cats. The implanted coil was inductively coupled to an external 12 x 12 cm square surface coil that was mounted on a 14-cm diameter Plexiglas cradle (Townsend Industries, Des Moines, IA). Two similar cradles were prepared with transmit-only 12 x 12 cm surface coils and either a receive-only 6 x 6 cm square surface coil or a receive-only quadrature coil pair (two 4 x 6 cm coils overlapped slightly to minimize their mutual inductance) with the same surface area (6 x 6 cm). A total of five single-slice, T1-weighted spin-echo images (TR = 500 ms, TE = 30 ms, 4-mm slice thickness) were acquired from a 1-liter saline phantom and from the second lumbar spinal level in an adult cat with a normal, uninjured spinal cord. On the spinal cord images, the quadrature coil exhibited a factor of 1.65 increase in SNR relative to the single-turn surface coil, whereas the implanted coil achieved a factor of 2.19 increase in SNR. The improved SNR for the quadrature and implanted coils was observed as a dramatic improvement in the clarity of the images.(ABSTRACT TRUNCATED AT 250 WORDS)