We examine surveys of M dwarfs within 20 pc to determine the incidence of stellar companions. Observational data are drawn from high-quality surveys, including IR-array imaging, precise velocities, IR speckle interferometry, and visual imaging, and their respective incompleteness is determined. Each technique permits detection of companions down to the H-burning limit, and each is nearly complete (owing to the proximity of M dwarfs) within a specific range of separation, the farthest being approximately 10(4) AU. The number of companions per primary per AU of semimajor axis, dN/da, is computed and found to decline monotonically toward larger semimajor axes. The period distribution (in bins of DELTA-log P) exhibits a unimodal, broad maximum with a peak in the range P = 9-220 yr, corresponding to separations 3-30 AU, similar to that for G dwarf binaries. Integrating over all semimajor axes yields the average number of companions per primary, 0.55 +/- 0.09, which includes those companions clumped in multiple systems. We determine the distribution of multiplicities (binary, triple, etc.) both empirically and hypothetically. The resulting multiplicity of M dwarf primaries is single:double:triple:quadruple = 58:33:7:1 for 100 primaries. Thus, 58% of nearby M dwarf primaries are single and 42% +/- 9% have companions, similar to Henry & McCarthy's binary frequency of 34%. The M-dwarf binary frequency is lower than that for G dwarfs (approximately 57%), owing partly to the smaller range of companion masses included, i.e., companions less massive than the M- or G-type primary. The mass function of companions to M dwarfs is roughly flat in shape, similar to the field mass function. Companions to G dwarfs also exhibit the field mass function, which provides support for protostellar "capture" as the dominant mechanism by which binaries form.
