An accurate model of synaptic integration in an individual neuron requires fidelity both in the time course of ynaptic input to the cell and in the dynamics of the cell's response to input. We have been working on both fronts to build a model of integration in visual cortical neurons. We have developed software that takes visual stimulus "movies" as input, uses published spatiotemporal response functions (either linear kernels or linear-nonlinear cascades) to calculate the response of afferent neurons, and generates spike trains that determine the timing of synaptic events at arbitrarily specified locations on a cell in the NEURON simulation environment. Multiple simultaneous input streams (e.g., both visual stimuli and top-down modulatory inputs) are supported. This stimulus emulation package has been used to model sensory and attentional input to individual neurons in the primate visual cortex, but it is easily adapted to any region where the topographical organization of receptive field properties and the time course of activity have been characterized. We have also been using results from published voltage-clamp studies of ion channels in neocortical pyramidal neurons, together with recordings from slice and intact preparations, to constrain an improved compartmental model. These models support detailed study of mechanisms of synaptic integration by allowing direct comparison of results from physiological experiments to a biophysically detailed model receiving similar input, and are publicly available.