The Met receptor tyrosine kinase and its ligand hepatocyte growth factor (HGF) play an important role in mediating both tumor progression and tissue regeneration. The N-terminal and first Kringle domains (NK1) of HGF comprise a naturally occurring splice variant that retains the ability to activate the Met receptor. However, NK1 is a weak agonist and is relatively unstable, limiting its therapeutic potential. Here, we engineered NK1 mutants with improved biochemical and biophysical properties that function as Met receptor agonists or antagonists. We first engineered NK1 for increased stability and recombinant expression yield using directed evolution. The NK1 variants isolated from our library screens acted as weak Met receptor antagonists due to a mutation at the NK1 homodimerization interface. We introduced point mutations that restored this NK1 homodimerization interface to create an agonistic ligand, or that further disrupted this interface to create more effective antagonists. The rationally engineered antagonists exhibited melting temperatures up to approximately 64 °C, a 15 °C improvement over antagonists derived from wild-type NK1, and approximately 40-fold improvement in expression yield. Next, we created disulfide-linked NK1 homodimers through introduction of an N-terminal cysteine residue. These covalent dimers exhibited nearly an order of magnitude improved agonistic activity compared to wild-type NK1, approaching the activity of full-length HGF. Moreover, covalent NK1 dimers formed from agonistic or antagonistic monomeric subunits elicited similar activity, further signifying that NK1 dimerization mediates agonistic activity. These engineered NK1 proteins are promising candidates for therapeutic development and will be useful tools for further exploring determinants of Met receptor activation.