A Rhesus monkey was trained to discriminate between 2 acoustic signals, preceded by visual cues, that instructed which of 2 movements to make. In 75% of the trials the visual cue correctly indicated which of the 2 acoustic signals would follow. In 25% of the trials the visual cue was mislcading, readying for the wrong acoustic signal. Based on the visual cue the monkey could anticipate which acoustic signal would be presented, a possibility confirmed by behavioural analysis. Movements conforming with the acoustic instruction signals were reinforced. The activity of 65 medial geniculate nucleus (MGN) and 107 primary auditory cortex (ACx) neurons was recorded during task performance. Significant response differences to anticipated vs unanticipated acoustic signals were found in 32% of the MGN and 27% of the ACx neurons. In both structures some of the affected neurons responded more vigorously to the correctly anticipated acoustic signals while others responded more vigorously to the unanticipated ones.An index of relative response intensity (RRI) was derived for each neuron by comparing its response to the sound signal in reinforced trials with that obtained in trials with no visual cues and no reinforcement (a 'blank' condition). For both MGN and ACx the analysis of RRIs revealed two anticipation-related effects. One consisted of response facilitation to anticipated sound signals together with suppression of responses to unanticipated sound signals. The other consisted of facilitation of responses to unanticipated sound signals, combined with base-line responsiveness to anticipated sounds. These effects can be interpreted to reflect a context-relevant trial by trial selection of neuronal channels which conduct information along the thalamo-cortical segment of the auditory system. The similarity between the effects of anticipation at the thalamic and the cortical levels suggests that the changes in excitability which underlay trial by trial neuronal selectivity occur subcortically.