The Palos Verdes peninsula is a short, very narrow (< 3 km) shelf in southern California that is bracketed by two large embayments. In May 1992, arrays of up to 4 moorings and 2 benthic tripods were deployed in a yearlong study of the circulation processes over this shelf and the adjacent slope. Wind stress, coastal sea level, atmospheric pressure and wave records were obtained from offshore sites and from coastal stations surrounding Palos Verdes. Bottom stress calculated for the mid-shelf sites using a boundary-layer model and data from the above instruments indicated the bottom drag coefficient over this shelf is about 0.003 Currents flow toward the northwest along the shelf and upper slope. Speeds are generally around 20-30 cm/s. There was no obvious seasonal structure in the flow. The first EOF for subtidal alongshelf current accounted for nearly 70% of the variance at sites on the shelf and upper slope. The dominant fluctuations had periods between 5 and 20 days, periods longer than seen in the regional wind stress field. Coastal sea level and the alongshore gradient in sea level had a similar concentration of energy in the 5-20 day frequency band. About 30% of the alongshelf flow was coherent with the alongshelf pressure gradient; currents flowed down the pressure gradient with minimal phase lag. Winds accounted for only 15-20% of the variance in subtidal currents, but the measured effect of wind stress was large. A 1 dyne/cm2 wind stress was associated with a 20-30 cm/s alongshore current. Both the regional wind stress and the alongshelf pressure gradients had spatial scales much larger than found on this small shelf. Subtidal flows forced by these regional fields were set up in the adjacent, much broader basins. The currents amplified as they moved onto the narrow shelf between the basins. Hence, local wind-driven currents had anomalously large amplitudes. The momentum equations for alongshelf wind or pressure gradients did not balance because some of the measured terms were associated with regional fields, others with local process. Our observations suggest that it is more difficult to determine which measured fields reflect the local processes in regions with rapidly changing topography. ?? 2002 Elsevier Science Ltd. All rights reserved.