We study a sample of about 1400 disk M dwarfs that are found in 148 fields observed with the Wide Field Camera 2 (WFC2) on the Hubble Space Telescope and 162 fields observed with pre-repair Planetary Camera 1 (PC1), of which 95 of the WFC2 fields are newly analyzed. The method of maximum likelihood is applied to derive the luminosity function and the Galactic disk parameters. At first, we use a local color-magnitude relation and a locally determined mass-luminosity relation in our analysis. The results are consistent with those of previous work but with considerably reduced statistical errors. These small statistical errors motivate us to investigate the systematic uncertainties. Considering the metallicity gradient above the Galactic plane, we introduce a modified color-magnitude relation that is a function of Galactic height. The resultant M dwarf luminosity function has a shape similar to that derived using the local color-magnitude relation but with a higher peak value. The peak occurs at $M_V \sim 12$ and the luminosity function drops sharply toward $M_V \sim 14$. We then apply a height-dependent mass-luminosity function interpolated from theoretical models with different metallicities to calculate the mass function. Unlike the mass function obtained using local relations, which has a power-law index $��= 0.47$, the one derived from the height-dependent relations tends to be flat ($��= -0.10$). The resultant local surface density of disk M dwarfs (12.2 +/- 1.6 M_sun pc^{-2}) is somewhat smaller than the one obtained using local relations (14.3 +/- 1.3 M_sun pc^{-2}). Our measurement favors a short disk scale length, H = 2.75 +/- 0.16 (statistical) +/- 0.25 (systematic) kpc.

20 pages, 10 ps figures, accepted for publication in ApJ