Previous work suggested that the cerebellum and red nucleus are not necessary for the acquisition, extinction, and reacquistion of the in vitro classically conditioned abducens nerve response in the turtle. These findings are extended in the present study by obtaining conditioned responses (CRs) in preparations that received a partial ablation of the brain stem circuitry. In addition to removing all tissue rostral to and including the midbrain and cerebellum, a transection was made just caudal to the emergence of the IXth nerve. Such ablations result in a 4-mm-thick section of brain stem tissue that functionally eliminates the sustained component of the unconditioned response (UR) while leaving only a phasic component. We refer to this region of brain stem tissue caudal to the IXth nerve as the "caudal premotor blink region." Neural discharge was recorded from the abducens nerve following a single shock unconditioned stimulus (US) applied to the ipsilateral trigeminal nerve. When the US was paired with a conditioned stimulus (CS) applied to the posterior eighth, or auditory, nerve using a delay conditioning paradigm, a positive slope of CR acquisition was recorded in the abducens nerve, and CR extinction was recorded when the stimuli were alternated. Resumption of paired stimuli resulted in reacquisition. Quantitative analysis of the CRs in preparations in which the caudal premotor blink region had been removed and those with cerebellar/red nucleus lesions showed that both types of preparations had abnormally short latency CR onsets compared with preparations in which these regions were intact. Preparations with brain stem transections had significantly earlier CR offsets as more CRs terminated as short bursts when compared with intact or cerebellar lesioned preparations. These data suggest that a highly reduced in vitro brain stem preparation from the turtle can be classically conditioned. Furthermore, the caudal brain stem is not a site of acquisition in this reduced preparation, but it contributes to the sustained activity of both the UR and CR. Finally, the unusually short CR onset latencies following lesions to the cerebellum are not further exacerbated by removal of the caudal brain stem. These studies suggest that convergence of CS and US synaptic inputs onto the abducens nerve reflex circuitry may underlie acquisition in this reduced preparation, but that mechanisms that control learned CR timing arise from the cerebellorubral system.