Power spectra of daily mean surface air temperature at eight locations in the United States were computed for each season of a 21-year period (March 1943–February 1969). After partitioning the spectra into what appear to be physically meaningful frequency bands, time series of power in spectral segments were related statistically to time series of seasonal mean 700-mb heights at each point of a grid extending from mid-Atlantic to mid-Pacific. Gross aspects of resultant fields of simple linear correlation coefficients were found amenable to reasonable physical interpretation. In an overall sense, power is concentrated in progressively longer period temperature oscillations as the seasonal mean circulation proceeds from fast westerlies in the vicinity of the station to highly amplified wave patterns. Certain previously established persistent circulation regimes are shown by the present study to be associated with a concentration of power in the long-period portion of the power spectrum of daily temperature. Multiple linear regression equations were derived relating the seasonal mean 700-mb height field to the apportionment of power in the variance spectra of daily mean temperatures. Validation of these equations awaits the accumulation of an adequate independent data sample.