Highly active nickel phosphide (Ni₂P) nanoclusters confined in a mesoporous SiO₂ catalyst were synthesized by a two-step process targeting tight control over the Ni₂P size and phase. The Ni precursor was incorporated into the MCM-41 matrix by one-pot synthesis, followed by the phosphorization step, which was accomplished in oleylamine with trioctylphosphine at 300 °C so to achieve the phase transformation from Ni to Ni₂P. For benchmarking, Ni confined by the mesoporous SiO₂ (absence of phosphorization) and 11 nm Ni₂P nanoparticles (absence of SiO₂) was also prepared. From the microstructural analysis, it was found that the growth of Ni₂P nanoclusters was restricted by the mesoporous channels, thus forming ultrafine and highly dispersed Ni₂P nanoclusters (<2 nm). The above approach led to promising catalytic performance following the order u-Ni₂P@m-SiO₂ > n-Ni₂P > u-Ni@m-SiO₂ > c-Ni₂P in the selective hydrogenation of SO₂ to S. In particular, u-Ni₂P@m-SiO₂ exhibited SO₂ conversions of 94% at 220 °C and ∼99% at 240 °C, which are higher than the 11 nm stand-alone Ni₂P particles (43% at 220 °C and 94% at 320 °C), highlighting the importance of the role played by SiO₂ in stabilizing ultrafine nanoparticles of Ni₂P. The reaction activation energy E_a over u-Ni₂P@m-SiO₂ is ∼33 kJ/mol, which is lower than those over n-Ni₂P (∼36 kJ/mol) and c-Ni₂P (∼66 kJ/mol), suggesting that the reaction becomes energetically favored over the ultrafine Ni₂P nanoclusters.