Modulation of receptive field properties of thalamic somatosensory neurons by the depth of anesthesia. The dominant frequency of electrocorticographic (ECoG) recordings was used to determine the depth of halothane or urethan anesthesia while recording extracellular single-unit responses from thalamic ventral posterior medial (VPM) neurons. A piezoelectric stimulator was used to deflect individual whiskers to assess the peak onset latency, magnitude, probability of response, and receptive field (RF) size. There was a predictable increase in the dominant ECoG frequency from deep stage IV to light stage III-1 anesthetic levels. There was no detectable frequency at stage IV, a 1- to 2-Hz dominant frequency at stage III-4, 3-4 Hz at stage III-3, 5-7 Hz at stage III-2, and a dual 6- and 10- to 13-Hz pattern at stage III-1. Reflexes and other physical signs showed a correlation with depth of anesthesia but exhibited too much overlap between stages to be used as a criterion for any single stage. RF size and peak onset latency of VPM neurons to whisker stimulations increased between stage III-4 and III-1. A dramatic increase in RF size and response latency occurred at the transition from stage III-3 (RF size approximately 2 whiskers, latency approximately 7 ms) to stage III-2 (RF size approximately 6 whiskers, latency approximately 11 ms). Response probability and magnitude decreased from stage III-4 to stage III-3 and III-2. No responses were ever evoked in VPM cells by vibrissa movement at stage IV. These changes in VPM responses as a function of anesthetic depth were seen only when the nucleus principalis (PrV) and nucleus interpolaris (SpVi) trigeminothalamic pathways were both intact. Eliminating SpVi inputs to VPM, either by cutting the primary trigeminal afferent fibers to SpVi or cutting axons projecting from SpVi to VPM, immediately reduced the RF size to fewer than three whiskers. In addition, the predictable changes in VPM response probability, response magnitude, and peak onset latency at different anesthetic depths were all absent after SpVi pathway interruption. We conclude that 1) the PrV input mediates the near "one-to-one" correspondence between a neuronal response in VPM and a single mystacial whisker, 2) in contrast, the SpVi input to VPM is primarily responsible for the RF properties of VPM neurons at light levels of anesthesia and presumably in the awake animal, and 3) alterations in VPM responses produced by changing the depth of anesthesia are due to its selective influence on the properties mediated by SpVi inputs at the level of the thalamus.