Fine-tuning of temporally precise behaviours by the cerebellum requires diverse cell types and specific connectivity to generate the neural activity underlying sensory-motor transformations. Molecular layer interneurons (MLIs) are well-positioned to influence cerebellar cortical output activity by Purkinje cells and gate synaptic plasticity underlying motor adaptation. However, how the diversity of MLI anatomy, intrinsic properties and connectivity influence cerebellar cortical circuit computations is not understood. We hypothesize that specific lamina-dependent properties of MLI circuits are a critical mechanism of functional MLI diversity. To address this question, we monitored MLI activity using high-speed two-photon calcium imaging in awake mice, in response to air stimuli of the face and whiskers. Brief puff stimuli to the face evoked Ca2+ responses in deep MLIs displayed gradually smaller amplitudes and on average 10 ms faster response onsets as compared with those in the outer molecular layer. We identified a transcriptional cluster of MLIs (Penk1) that marked putative basket cells to examine if MLI classes could account for laminar dependence of MLI responses. We did not observe a difference in activity response kinetics in Penk1+ MLIs. In contrast, we did observe a laminar difference of onset delays of granule cell axonal responses, suggesting an input-inherited spatial dependence of MLIs responses. In apparent contrast to previous studies, the peak amplitude of all MLI responses did not correlate with the peak whisker deflection angle. However, for a subset of MLIs in which their Ca2+ response onset occurred after the initiation of whisking, response amplitudes correlated with the whisker deflection. These slow-onset MLIs were located preferentially in the outer molecular layer. Longer stimulation induced prolonged Ca2+ activity in over half of the cells in the inner, middle and outer molecular layers, suggesting a robust somatosensory representation throughout the molecular layer. During spontaneous whisking, Ca2+ response amplitudes correlated with the whisker deflection angle, suggesting that somatosensory response occludes motor-related responses during air puff stimuli. Thus, our results are consistent with a layer-specific sensorimotor representation in MLIs that is, at least in part, dictated by the intrinsic GC input properties and/or a spatial segregation of whisk-related information across GC inputs.