We present our analysis of the EXOSAT medium energy (ME) archival data on dwarf novae. Many dwarf novae were observed with EXOSAT; unfortunately, a significant fraction resulted in nondetection or marginal detection with the ME instrument. Our analysis of the brightest dwarf nova, SS Cyg, shows little correlation between the count rate and the hardness ratio, contrary to an earlier report. This is likely to be the result of a slight error in background subtraction performed. The apparent lack of correlation favors a compact boundary layer between the accretion disk and the white dwarf as the hard X-ray emitting region, rather than an extended ''corona.'' According to our spectral fit, all dwarf nova data are compatible with a single-temperature (2->20 keV) bremsstrahlung continuum with an iron emission line near 6.7 keV. However, several sources show significant epoch-to-epoch variability, the origin of which is poorly understood at the moment. Better hard X-ray data are needed to advance our understanding of dwarf novae. With this knowledge, we investigate the contribution of dwarf novae to the Galactic ridge X-ray emission. They are found to be spectrally similar; using a reasonable estimate of the spatial distribution of dwarf novae, these systems appear to be a major, if not the dominant, component of the ridge emission.