Underwater　images　often　suffer　from　serious　degradations　in　view　of　complex　and　changeable　marine　environment.　The　current　quality　improvement　strategies　for　degraded　underwater　images　can　be　divided　into　enhancement　and　restoration　methods　according　to　whether　they　are　based　on　degrading　models.　However,　the　existing　algorithms　often　miss　the　distorted　area　or　have　poor　performance　when　dealing　with　multi-type　distorted　images.　To　overcome　the　above　limitation,　we　propose　a　Distortion-Guided　strategy　for　efficient　Underwater　Image　Restoration　(DGUIR).　DGUIR　splits　the　restoration　task　into　two　stages:　distorted　area　detection　and　targeted　restoration.　In　addition,　the　knowledge　generated　in　the　distortion　detection　stage　will　guide　the　subsequent　targeted　restoration.　It　is　worth　mentioning　that　DGUIR　is　able　to　address　multiple　types　of　distortions　with　only　one　model,　and　exhibits　superior　performance　in　comparison　with　existing　algorithms.