Multiphoton excitation of molecular probes has become an important tool in experimental neurobiology owing to the intrinsic optical sectioning and low light scattering it affords. Using molecular functional indicators, multiphoton excitation allows physiological signals within single neurons to be observed from within living brain tissue. Ideally, it would be possible to record from multiple sites located throughout the elaborately branching dendritic arbors, in order to study the correlations of structure and function both within and across experiments. However, existing multiphoton microscope systems based on scanning mirrors do not allow optical recordings to be obtained from more than a handful of sites simultaneously at the high rates required to capture the fast physiological signals of interest (>100Hz for Ca2+ signals, >1kHz for membrane potential transients). In order to overcome this limitation, two-dimensional acousto-optic deflection was employed, to allow an ultrafast laser beam suited for multiphoton excitation to be rapidly repositioned with low latency (∼15mus). This supports a random-access scanning mode in which the beam can repeatedly visit a succession of user-selected sites of interest within the microscope's field-of-view at high rates, with minimal sacrifice of pixel dwell time. This technique of acousto-optic multiphoton laser scanning microscope (AO-MPLSM) was demonstrated to allow the spatial profile of signals arising in response to physiological stimulation to be rapidly mapped. Means to compensate or avoid problems of dispersion which have hampered AO-MPLSM in the past are presented, with the latter being implemented. Separately, the combination of photon counting detection with multiphoton excitation, termed generally multiphoton photon counting spectroscopy (MP-PCS), was also considered, with particular emphasis on the technique of fluorescence correlation spectroscopy (FCS). MP-PCS was shown to allow information about molecular numbers and mobility, as well as the focal volume itself, to be obtained. This capability may in the future be employed to study the number and transport of native neuronal signaling molecules. MP-PCS was also found to be a promising off-line tool which can allow the performance of AO-MPLSM to be optimized, with respect to both the instrument and the indicators employed.