Abstract Due to their ubiquity and very long main-sequence lifetimes, M dwarfs provide an excellent tool to study the formation and chemical enrichment history of our Galaxy. However, owing to their intrinsic faintness, the acquisition of high-resolution, high signal-to-noise spectra of low-mass stars has been limited to small numbers of very nearby stars, mostly from the Galactic disk population. On the other hand, large numbers of low- to medium-resolution spectra of M-type dwarf stars from both the local Galactic disk and halo are available from various surveys. In order to fully exploit these data, we develop a template-fit method using a set of empirically assembled M dwarf/subdwarf classification templates, based on the measurements of the TiO and CaH molecular bands near 7000 Å, which are used to classify M dwarfs/subdwarfs by spectral type and metallicity class. We further present a pipeline to automatically determine the effective temperature T eff, metallicity [M/H], α-element to iron abundance ratio [α/Fe], and surface gravity log g of M dwarfs/subdwarfs using the latest version of BT-Settl model atmospheres. We apply these methods to a set of low- to medium-resolution spectra of 1544 high proper-motion (μ ≥ 0.″4 yr−1) M dwarfs/subdwarfs, collected at the MDM observatory, Lick Observatory, Kitt-Peak National Observatory, and Cerro-Tololo Interamerican Observatory. Our metallicity estimates appear to be consistent with the expected color–magnitude variation of stars relative to the atmospheric composition, as our sample shows a clear stratification with respect to metallicity in the Hertzsprung–Russel diagram constructed from their Gaia DR2 parallaxes and optical magnitudes. Furthermore, the measured chemical parameters of the two components in 48 binary systems are in good agreement with each other, which suggest a precision of ±0.22 dex in [M/H], ±0.08 dex in [α/Fe], and ±0.16 dex in the combined index [α/Fe] + [M/H]. We find that the relationship between color and spectral subtype depends on metallicity class, as the color G BP − G RP is more sensitive to subtype for metal-rich M dwarfs in comparison to metal-poor M subdwarfs. We also demonstrate that effective temperature as a function of spectral subtype has a steeper slope for metal-rich M dwarfs than metal-poor M subdwarfs. There is also a good consistency between “metallicity class,” obtained from the empirical classification templates, and the index [α/Fe] + [M/H] (∼[α/H]), obtained from BT-Settl model fitting, which means that the more easily measured “metallicity class” can be used as a relatively reliable indicator of absolute α-element abundance, [α/H], in low-mass stars. Finally, we examine the distribution of our stars in the [α/Fe] versus [M/H] diagram, which shows evidence of clustering in chemical abundance makeup, suggestive of discrete populations among the local disk and halo stars. We predict that analyses of larger samples of spectra of nearby M-type stars will uncover a complex structure of our Galaxy.