We present a stochastic model of midlatitude sea surface temperature (SST) anomalies that includes the gustiness of sea surface winds, and examine the effects on atmosphere-ocean thermal coupling. Atmosphere-ocean coupling depends on conditions in the respective boundary layers. In many applications, including the common stochastic view of midlatitude SST anomaly variability as the response to white-noise weather fluctuations, a constant, averaged, or too slowly varying wind speed is used in standard heat flux bulk formulae, and the (effectively stochastic) gustiness of the winds is not taken into account. Nevertheless, observations at Ocean Weather Ship P (and other stations) reveal that stochastic wind fluctuations are crucial to explain the evolution of SST anomalies: the predictions of an extended version of the stochastic SST anomaly model in which the feedback coefficient is not constant but instead includes a stochastic process are in close agreement with observations. For example, our model (counterintuitively) predicts that atmospheric noise can significantly increase the persistence of midlatitude SST anomalies. Changes in the high-frequency atmospheric variability associated with ENSO may also impact midlatitude SST variability.