diff --git a/src/xenia/cpu/backend/x64/x64_sequences.cc b/src/xenia/cpu/backend/x64/x64_sequences.cc
index 5ee8bed75..b4593f222 100644
--- a/src/xenia/cpu/backend/x64/x64_sequences.cc
+++ b/src/xenia/cpu/backend/x64/x64_sequences.cc
@@ -4493,12 +4493,39 @@ EMITTER(SHL_I64, MATCH(I<OPCODE_SHL, I64<>, I64<>, I8<>>)) {
     EmitShlXX<SHL_I64, Reg64>(e, i);
   }
 };
+EMITTER(SHL_V128, MATCH(I<OPCODE_SHL, V128<>, V128<>, I8<>>)) {
+  static void Emit(X64Emitter& e, const EmitArgType& i) {
+    // TODO(benvanik): native version (with shift magic).
+    if (i.src2.is_constant) {
+      e.mov(e.r9, i.src2.constant());
+    } else {
+      e.mov(e.r9, i.src2);
+    }
+    e.lea(e.r8, e.StashXmm(0, i.src1));
+    e.CallNativeSafe(reinterpret_cast<void*>(EmulateShlV128));
+    e.vmovaps(i.dest, e.xmm0);
+  }
+  static __m128i EmulateShlV128(void*, __m128i src1, uint8_t src2) {
+    // Almost all instances are shamt = 1, but non-constant.
+    // shamt is [0,7]
+    uint8_t shamt = src2 & 0x7;
+    alignas(16) vec128_t value;
+    _mm_store_si128(reinterpret_cast<__m128i*>(&value), src1);
+    for (int i = 0; i < 15; ++i) {
+      value.u8[i ^ 0x3] = (value.u8[i ^ 0x3] << shamt) |
+                          (value.u8[(i + 1) ^ 0x3] >> (8 - shamt));
+    }
+    value.u8[15 ^ 0x3] = value.u8[15 ^ 0x3] << shamt;
+    return _mm_load_si128(reinterpret_cast<__m128i*>(&value));
+  }
+};
 EMITTER_OPCODE_TABLE(
     OPCODE_SHL,
     SHL_I8,
     SHL_I16,
     SHL_I32,
-    SHL_I64);
+    SHL_I64,
+    SHL_V128);
 
 
 // ============================================================================