Purpose: Bennett and Sekuler (1992) showed thai the visual system performs a spatial analysis of the luminance profile of two-component sinusoidal gratings. How long does it take the visual system to construct accurate representations of spatial profiles? The present experiments attempted to determine if spatial profile analysis is a rapid or relatively slow process. Methods: The stimuli were f+2f compound gratings. The target stimuli contained a contrast increment to 2f. In fixed conditions, overall contrast and frequency remained constant across trials. In profile conditions, overall contrast and frequency was varied by ±0.3 log units across trials. Expt. I used a 2-IFC procedure to measure discrimination thresholds for stimulus durations ranging from H to 112 ms. Expt. 2 used a speeded yes-no response procedure: observers were forced to respond within a certain temporal interval (ranging from unlimited (no deadline) to 300 ms). Rcjults: Similar results were found in both experiments. Thresholds increased as stimulus duration decreased and accuracy declined as response deadline was shortened at the same rates in fixed and profile conditions. In both experiments we obtained reliable profile effects at all but the shortest duration and deadline tested. Profile effects were found with stimulus exposure as short as 28 ms and a deadline as short as 400 ms. Conclusions: Our data suggest that constructing accurate representations of spatial profiles is not a slow process; rather, spatial profiles are constructed very rapidly. However, our data do not allow us to specify the time course precisely. We are examining this issue further by measuring discrimination thresholds for patterns whose overall contrast are.