The heat budget of a cross-shelf section extending 16 km offshore of the outer banks of North Carolina is studied during two time periods: August 1994 and October 1994, using data collected as part of the Coastal Ocean Processes (CoOP) Inner Shelf Study. Heat budgets are computed on two different timescales: monthly averages over August and October, which reflect seasonal variations, and a fluctuation budget, which reflects variation on daily to weekly timescales. During August, a period of strong stratification, the increase in the area-averaged water temperature (approximately 3.2 ¿C) was due primarily to the surface heat flux. Fluctuations in temperature during August were caused primarily by the cross-shelf heat flux, due to wind-driven upwelling and downwelling circulation. In October, the area-averaged shelf temperature dropped by approximately 3.5 ¿C due to both surface heat loss and the alongshore transport of heat. Weak vertical stratification in October led to small cross-shelf heat fluxes, and temperature fluctuations in October were due primarily to fluctuations in the surface and alongshore heat fluxes. In both August and October, variation on daily to weekly timescales of the area-averaged temperature of the shelf was strongly correlated with the alongshore component of the wind stress. In August, alongshore poleward winds caused upwelling and the area-averaged temperature decreased; conversely, equatorward winds caused downwelling and warming. In October, although the variations in temperature were smaller, alongshore winds were positively correlated with alongshore currents and the surface heat flux (for reasons discussed by Austin and Lentz ), so that poleward winds resulted in warming; conversely, equatorward winds resulted in cooling. Therefore the dependence of the change in heat content on the alongshore wind stress changed sign between August and October. A simple dynamical model was constructed to relate changes in heat content to the alongshore wind stress. The model results were compared to 12 years of meteorological records from the Coastal Engineering Research Center's Field Research Facility, directly onshore of the experimental site. The results suggest a seasonal cycle in the dominant fluctuating heat balance, consistent with the field results found for August and October 1994. In May through August, cross-shelf flux dominates variation in the heat content. In October through March, the surface heat flux and alongshore heat flux dominate the variation. ¿ 1999 American Geophysical Union