In　this　paper,　we　consider　the　choice　of　pilot　estimators　for　the　single-index　varying-coefficient　model,　which　may　result　in　radically　different　estimators,　and　develop　the　method　for　estimating　the　unknown　parameter　in　this　model.　To　estimate　the　unknown　parameters　efficiently,　we　use　the　outer　product　of　gradient　method　to　find　the　consistent　initial　estimators　for　interest　parameters,　and　then　adopt　the　refined　estimation　method　to　improve　the　efficiency,　which　is　similar　to　the　refined　minimum　average　variance　estimation　method.　An　algorithm　is　proposed　to　estimate　the　model　directly.　Asymptotic　properties　for　the　proposed　estimation　procedure　have　been　established.　The　bandwidth　selection　problem　is　also　considered.　Simulation　studies　are　carried　out　to　assess　the　finite　sample　performance　of　the　proposed　estimators,　and　efficiency　comparisons　between　the　estimation　methods　are　made.