Moored time series from the Coastal Ocean Dynamics Experiment (CODE), Shelf Mixed Layer Experiment (SMILE), Sediment Transport Events over the Shelf and Slope (STRESS) study, and Northern California Coastal Circulation Study (NCCCS) are used to study subtidal cross-shelf circulation over the northern California shelf. The northern California shelf, like much of the United States Pacific coast, is subject to strong wind forcing which exhibits characteristic seasonality. In winter and early spring, it is distinguished by poleward and equatorward fluctuations on time scales of days and by weak monthly means. In summer, it is distinguished by periods of equatorward stress lasting several weeks and by relatively strong monthly means. The intensive winter and spring SMILE and STRESS and summer CODE-2 field programs permit the examination of cross-shelf circulation under both types of wind forcing conditions at a mid-shelf site (~90 m) 6 km from the northern California coast. The primary thesis goal is to examine the applicability of a two-dimensional conceptual model of wind-forced cross-shelf circulation. In this conceptual model, surface and bottom cross-shelf flows are forced by along-shelf wind stress and bottom stress, and interior cross-shelf flow compensates such that the depth-averaged flow is zero. A secondary thesis goal is to use the seasonal coverage of available field programs to gain insight into seasonal variability of cross-shelf circulation on the northern California shelf. To accomplish these goals, the observed subtidal cross-shelf circulation is examined in the context of the winter and spring heat and salt balances, an analytic model of wind-forced cross-shelf circulation, and the spatial scales of subtidal velocity. Mean and fluctuating heat and salt balances estimated between December, 1988 and May, 1989 demonstrate the importance of cross-shelf fluxes and their general consistency with the simple conceptual model. Mean fluxes are consistent with the weak mean equatorward wind stress observed during SMILE. The dominant terms in the fluctuating balances are the cross-shelf fluxes and local changes in heat and salt content. These are well correlated with each other and with the local along-shelf wind stress. The along-shelf heat flux divergence is of secondary importance to the fluctuating heat balance. It is uncorrelated with the along-shelf wind stress, and occurrences when it is strong are interpreted as effects of mesoscale features. To examine the applicability of the wind-forced conceptual model in more detail, a simple analytic model incorporating the assumptions of the conceptual model and observed local wind forcing is compared quantitatively to estimates of surface mixed layer, interior, and bottom mixed layer cross-shelf transport for winter SMILE and STRESS and summer CODE-2 observations. This comparison suggests the model is more suited to the transient wind forcing observed during SMILE and STRESS than to the steady wind forcing observed during CODE-2. For 2-3 day wind events between December, 1988 and February, 1989, the model is well correlated with observed depthdependent (total minus depth-averaged) transports throughout the water column and with total surface mixed layer transports. For 2-3 week wind events between April and July, 1982, the model does not work nearly as well below the surface mixed layer. In the absence of other processes, the locally wind-forced model implies that the wind stress sets the horizontal scales of subtidal velocity. Correlation scales estimated for subtidal along-shelf velocity over the northern California shelf are for all field programs longer than the maximum mooring separation (60 km) and are similar to those of the wind stress. However, along-shelf correlation scales of cross-shelf velocity are shorter than minimum mooring separations for CODE. SMILE and NCCCS time series do resolve along-shelf correlation scales for near surface cross-shelf velocity. During this time, along-shelf correlation scales for near surface cross-shelf velocity vary on a monthly time scale. They are generally long (30 km or more) when correlation with wind stress is high and short (15 km or less) when correlation with wind stress is low. On at least one occasion, short along-shelf correlation scales coincide with the intrusion of an offshore mesoscale feature onto the shelf. Results of the three studies show the two-dimensional model offers some insight into the observed subtidal cross-shelf circulation, particularly in winter. During this time, the heat balance, analytical transport model, and correlation scales all provide evidence that the winter wind-forced circulation is quasi-two-dimensional. Threedimensional variability on the shelf, though important on occasion, does not appear to be wind-driven and may result from the influence of offshore mesoscale features. A quite different story emerges for summer when the simple conceptual model of crossshelf circulation fails to describe adequately subsurface cross-shelf flow. Two useful areas of further investigation may be the non-linear response of cross-shelf velocity to wind forcing and its response to other processes such as remotely generated mesoscale features.