Amyotrophic lateral sclerosis (ALS) is a fatal, adult-onset degenerative disorder of motor neurons. The diseased spinal cord motor neurons of more than 95% of amyotrophic lateral sclerosis (ALS) patients are characterized by the mis-metabolism of the RNA/DNA-binding protein TDP-43 (ALS-TDP), in particular, the presence of cytosolic aggregates of the protein. Most available mouse models for the basic or translational studies of ALS-TDP are based on transgenic overexpression of the TDP-43 protein. Here, we report the generation and characterization of mouse lines bearing homologous knock-in of fALS-associated mutation A315T and sALS-associated mutation N390D, respectively. Remarkably, the heterozygous TDP-43 (N390D/+) mice but not those heterozygous for the TDP-43 (A315T/+) mice develop a full spectrum of ALS-TDP-like pathologies at the molecular, cellular and behavioral levels. Comparative analysis of the mutant mice and spinal cord motor neurons (MN) derived from their embryonic stem (ES) cells demonstrates that different ALS-associated TDP-43 mutations possess critical ALS-causing capabilities and pathogenic pathways, likely modified by their genetic background and the environmental factors. Mechanistically, we identify aberrant RNA splicing of spinal cord Bcl-2 pre-mRNA and consequent increase of a negative regulator of autophagy, Bcl-2, which correlate with and are caused by a progressive increase of TDP-43, one of the early events associated with ALS-TDP pathogenesis, in the spinal cord of TDP-43 (N390D/+) mice and spinal cord MN derived from their ES cells. The TDP-43 (N390D/+) knock-in mice appear to be an ideal rodent model for basic as well as translational studies of ALS-TDP.