Population calcium signals generated by the action potential activity of local clusters of neurons have been recorded in the auditory cortex of mice using an optical fibre-based approach. These network calcium transients (NCaTs) occurred spontaneously as well as in response to sound stimulation. Two-photon calcium imaging experiments suggest that neurons and neuropil contribute about equally to the NCaT. Sound-evoked calcium signals had two components: an early, fast increase in calcium concentration, which corresponds to the short-latency spiking responses observed in electrophysiological experiments, and a late, slow calcium transient which lasted for at least 1 s. The slow calcium transients evoked by sound were essentially identical to spontaneous NCaTs. Their sizes were dependent on the spontaneous activity level at sound onset, suggesting that spontaneous and sensory-evoked NCaTs excluded each other. When using pure tones as stimulus, the early evoked calcium transients were more narrowly tuned than the slow NCaTs. The slow NCaTs were correlated with global ‘up states’ recorded with epidural potentials, and sound presented during an epidural ‘down state’ triggered a calcium transient that was associated with an epidural ‘up state’. Essentially indistinguishable calcium transients were evoked by optogenetic activation of local clusters of layer 5 pyramidal neurons in the auditory cortex, indicating that these neurons play an important role in the generation of the calcium signal. Taken together, our results identify sound-evoked slow NCaTs as an integral component of neuronal signalling in the mouse auditory cortex, reflecting the prolonged neuronal activity of local clusters of neurons that can be activated even by brief stimuli.