One important challenge in neuroscience is to examine the spiking activity of neuron assembly in a relevant physiological model. Of particular interest is the early olfactory circuit and the opportunity that such a well-defined circuit offers to decipher basic rules of sensory processing. Direct extra-cellular electrical measurements using multi-electrodes and spike-sorting algorithms have been used, mainly in insects, to examine odor-evoked neuronal activity. Such a technique allows simultaneous recordings of spatially compact groups of neurons with an excellent time resolution (< 10 µs), but without any cellular resolution, i.e. neurons are not individually identified. The understanding of some unresolved coding issues in olfaction, for example the manner in which odor representations are modified in space and time as they proceed from sensory neurons to their main target population will require recordings of neurons that are both large in number and that have been well-identified. Calcium imaging, using two-photon scanning microscopy has been proposed to examine such odor-evoked population activity and used in in vivo studies on insects and mammals. Nevertheless, the extraction of spiking activity from neuronal cytoplasmic calcium variation is not straightforward, since calcium signals are not coupled exclusively to spiking. We investigated this issue in locust projection neurons (PNs) by combining simultaneous intra-dendritic voltage recording and two-photon dendritic calcium imaging in vivo during odor stimulation in intact, non-anesthetized animals. Our goal was to determine whether dendritic calcium signals might be used to estimate a single PN spike output with reasonable accuracy in vivo.