The　advantage　of　FDS＇s　rich　dielectric　information　in　a　wide　band　is　made　full　use　for　applying　in　the　quantitative　evaluation　of　the　micro-water　content　of　oil-paper　insulation.　The　frequency　domain　dielectric　spectroscopy　of　oil-paper　insulation　samples　with　different　micro-water　content　is　analyzed　in　depth　based　on　the　principle　of　micro-physics.　Then,　the　backward　selection　multiple　regression　method　is　used　to　formulate　the　micro-water　content　evaluation　model.　The　significance　test　results　of　the　final　model　show　that　nine　extracted　characteristic　parameters　have　significant　influence　on　the　micro-water　content.　Only　the　established　multiple　regression　equation　at　the　test　frequency　of　10-1　Hz　is　linearly　significant.　At　the　end　of　the　article,　the　validity　of　the　model　is　tested　by　using　the　new　test　data.　The　relative　error　between　the　calculated　value　and　the　actual　value　is　mostly　less　than　4%　and　the　decision　coefficient　R2　of　the　equation　is　as　high　as　0.997　1.　The　reliability　of　the　model　is　fully　verified.　©　2020,　Science　Press.　All　right　reserved.