Abstract Fast optical signals can be used to study the time course of neuronal activity in localized cortical areas. The first report of such signals [Gratton, G., Corballis, P. M., Cho, E., Fabiani, M., Hood, D., 1995a. Shades of gray matter: Noninvasive optical images of human brain responses during visual stimulation. Psychophysiol, 32, 505-509.] was based on photon delay measures. Subsequently, other laboratories have also measured fast optical signals, but a debate still exists about how these signals are generated and optimally recorded. Here we report data from a visual stimulation paradigm in which different parameters (continuous: DC intensity; modulated: AC intensity and photon delay), wavelengths (shorter and longer than the hemoglobin isosbestic point), and source-detector distances (shorter and longer than 22.5 mm) were used to record fast signals. Results indicate that a localized fast signal (peak latency = 80 ms) can be detected with both delay and AC intensity measures in visual cortex, but not with unmodulated DC measures. This is likely due to the fact that differential measures (delay and AC intensity) are less sensitive to superficial noise sources, which heavily influence DC intensity. The fast effect had similar sign at wavelengths shorter and longer than the hemoglobin isosbestic point, consistent with light scattering but not rapid deoxygenation accounts of this phenomenon. Finally, the fast signal was only measured at source-detector distances greater than 22.5 mm, consistent with the intracranial origin of the signal, and providing indications about the minimum distance for recording. These data address some of the open questions in the field and provide indications about the optimal recording methods for fast optical signals.