Observed eddy heat fluxes, i.e., covariances between heat content and horizontal currents, are analyzed with moored data from the Coastal Ocean Dynamics Experiment. It is demonstrated that much of each sampled covariance arises from the separate correlations of currents and temperature with the upwelling variability, which is attributed to the particular warming and cooling processes present. Thus a large part of the observed eddy heat fluxes is wind driven. The remainder is consistent with being largely statistical noise in the observed covariance between those parts of the flow and temperature which are not correlated with the upwelling. This also suggests that the wind‐driven part alone might be a better estimate of the ensemble average eddy heat flu