The Polar Ultraviolet Imager (UVI) instrument can quantitatively determine important magnetospheric descriptors, notably including substorm onset time and global auroral output. Previous research has related the input variables of the magnetospheric system, namely solar wind parameters, to various output variables. However, a complex system such as the magnetosphere includes, in addition to inputs and outputs, state variables. Polar cap flux and magnetotail stretching are two such that can be estimated from the Defense Meteorological Satellite Program (DMSP) series satellites. We herein determine that both polar cap flux, &PHgr;c, and the magnetotail stretching index, b2i, do correlate well with 40-min averages of nightside auroral power observed by UVI. There were a total of n=638 distinct 40-min intervals within which b2i, &PHgr;c, and nightside auroral power could be determined. The correlations with premidnight auroral power were r=0.72 for &PHgr;c and r=-0.76 for b2i. The multiple-correlation coefficient of these two variables with nightside auroral power was 0.81. These sample correlations are far better than the sample correlations of solar wind input variables to nightside auroral power. Thus accurate space weather forecasting can demonstrably benefit greatly by monitoring current magnetospheric state variables (nowcasting), rather than attempting to reproduce output variables solely from solar wind inputs. Attempts to predict substorm onsets were less successful. Although the average polar cap flux prior to onset is larger than normal, the difference is not large enough to have significant predictive capability. Specifically, polar cap flux averaged 404¿133 and 422¿148 MWb for the entire years of 1996 and 1997, respectively, while the polar cap flux at the time of substorm onset averaged 455¿143 MWb. ¿ 2001 American Geophysical Union