Quality　Function　Deployment　(QFD)　is　a　systematic　approach　that　captures　customer　requirements　and　translates　them,　through　house　of　quality　(HOQ),　into　engineering　characteristics　of　product.　As　the　functional　relationships　between　customer　requirements　and　engineering　characteristics　in　QFD　are　uncertain,　unclear　and　fuzzy,　Radial　Basis　Function　(RBF)　to　determine　the　functional　relationships　for　QFD　is　presented,　and　a　QFD　functional　relationships　model　based　on　RBF　is　proposed.　According　to　RBF　neural　network　can　realize　the　nonlinear　mapping　space　from　the　input　space　to　the　output,　and　can　obtain　the　optimal　relationships　pattern　of　the　input　and　output,　the　customer　requirements　and　engineering　characteristics　in　QFD　constituted　the　input　and　output　of　the　RBF　Neural　Network　respectively,　the　optimal　relationships　are　constructed　through　the　neural　network　training.　A　case　study　of　natural　lighting　products　development　is　provided　to　illustrate　the　application　of　the　presented　method.