This　paper　is　concerned　with　the　statistical　inference　of　skew-normal　partial　functional　linear　models　which　give　a　useful　extension　of　the　normal　functional　regression　model.　The　maximum　likelihood　estimation　based　on　functional　principal　component　analysis　is　proposed.　Furthermore,　the　score　test　for　homogeneity　of　the　variance　is　discussed.　Asymptotic　properties　of　the　proposed　estimators　and　test　are　established　and　finite　sample　behavior　is　studied　through　a　small　simulation　experiment.　(C)　2019　Elsevier　B.V.　All　rights　reserved.