Empirical models of solar EUV irradiance variations developed by Hinteregger et al. (1981) and Tobiska and Barth (1990) are compared over time scales of the 27-day solar rotation and the 11-year solar cycle. Relative to the Hinteregger et al. model, the Tobiska and Barth model predicts the variations of the longest-wavelength EUV emissions (&lgr;>80 nm), which are entirely chromospheric, to be somewhat less during solar rotation and considerably less over the solar cycle. In contrast, emissions at the shortest EUV wavelengths (&lgr;<20 nm), which are predominantly of coronal origin, are predicted by the Tobiska and Barth model to vary considerably more during solar rotation and somewhat more during the solar cycle than by the Hinteregger et al. model. Within the wavelength region 20--80 nm, chromospheric emission variations are generally underpredicted on all time scales by the Tobiska and Barth model compared with that of Hinteregger et al., with differences between the predicted variations in the strongest chromospheric emission lines (He II 30.378 nm, He I 58.433 nm, and H I L&bgr; 102.572 nm exceeding a factor of 2. Neither models nor measurements yet provide a consistent picture of long-term variability in the EUV portion of the sun's spectrum. This is related to discrepancies of apparent instrumental origin which are identified between the rocket measurements of the EUV spectrum in August 1979 and in November 1988, to which the models are each independently tied. It is demonstrated that simple calculations of active region emission can provide useful constraints on empirical variability models, at least for chromospheric emissions.