We report here on measurements made from the 62nd story of the Williams Tower on the west side of Houston, Texas between 15 August and 15 September 2000. The time series of trace gases differ from those at many other urban sites in having very rapidly increasing spikes of O3, HCHO, and PAN. Measurements show that the highest O3 levels in Houston are not always those measured at the surface, and the extreme values may occur aloft. Plumes with high O3 appear to be produced largely from local sources and to have the potential to form additional O3. The ozone production efficiency (7 molecules of O3 produced per molecule of NOx consumed) when ΔO3/Δt ≥ 20 ppb per 15 min was found to be smaller than estimates made from observations directly downwind of the Ship Channel petrochemical plants (e.g., ~12). Back trajectories show that simple straight line flow was associated with mean O3 levels of 56 ppb, in contrast to flow patterns associated with a decrease in wind speed or flow reversal, which were associated with mean values of 63 ppb and extremes in excess of 125ppb. VOC samples taken during periods when ΔO3/Δt ≥ 20 ppb per 15 min were elevated and in particular light olefins were more than a factor of 7 greater than the corresponding samples collected on other occasions. No significant increase in isoprene at the Williams Tower was associated with these episodes. When air passed over stack emissions in eastern Houston, rich in VOCs, a Lagrangian model simulated O3 production rates of ~50 ppb hr-1.