diff --git a/llvm/lib/Target/AArch64/AArch64FrameLowering.cpp b/llvm/lib/Target/AArch64/AArch64FrameLowering.cpp
--- a/llvm/lib/Target/AArch64/AArch64FrameLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64FrameLowering.cpp
@@ -1985,13 +1985,14 @@
   bool EmitCFI = AFI->needsAsyncDwarfUnwindInfo(MF);
   bool HasWinCFI = false;
   bool IsFunclet = false;
-  auto WinCFI = make_scope_exit([&]() { assert(HasWinCFI == MF.hasWinCFI()); });
 
   if (MBB.end() != MBBI) {
     DL = MBBI->getDebugLoc();
     IsFunclet = isFuncletReturnInstr(*MBBI);
   }
 
+  MachineBasicBlock::iterator EpilogStartI = MBB.end();
+
   auto FinishingTouches = make_scope_exit([&]() {
     if (AFI->shouldSignReturnAddress(MF))
       authenticateLR(MF, MBB, NeedsWinCFI, &HasWinCFI);
@@ -1999,10 +2000,18 @@
       emitShadowCallStackEpilogue(*TII, MF, MBB, MBB.getFirstTerminator(), DL);
     if (EmitCFI)
       emitCalleeSavedGPRRestores(MBB, MBB.getFirstTerminator());
-    if (HasWinCFI)
+    if (HasWinCFI) {
       BuildMI(MBB, MBB.getFirstTerminator(), DL,
               TII->get(AArch64::SEH_EpilogEnd))
           .setMIFlag(MachineInstr::FrameDestroy);
+      if (!MF.hasWinCFI())
+        MF.setHasWinCFI(true);
+    }
+    if (NeedsWinCFI) {
+      assert(EpilogStartI != MBB.end());
+      if (!HasWinCFI)
+        MBB.erase(EpilogStartI);
+    }
   });
 
   int64_t NumBytes = IsFunclet ? getWinEHFuncletFrameSize(MF)
@@ -2040,7 +2049,7 @@
     // Adjust local stack
     emitFrameOffset(MBB, LastPopI, DL, AArch64::SP, AArch64::SP,
                     StackOffset::getFixed(AFI->getLocalStackSize()), TII,
-                    MachineInstr::FrameDestroy, false, NeedsWinCFI);
+                    MachineInstr::FrameDestroy, false, NeedsWinCFI, &HasWinCFI);
 
     // SP has been already adjusted while restoring callee save regs.
     // We've bailed-out the case with adjusting SP for arguments.
@@ -2092,16 +2101,17 @@
                                         NeedsWinCFI, &HasWinCFI);
   }
 
-  if (MF.hasWinCFI()) {
-    // If the prologue didn't contain any SEH opcodes and didn't set the
-    // MF.hasWinCFI() flag, assume the epilogue won't either, and skip the
-    // EpilogStart - to avoid generating CFI for functions that don't need it.
-    // (And as we didn't generate any prologue at all, it would be asymmetrical
-    // to the epilogue.) By the end of the function, we assert that
-    // HasWinCFI is equal to MF.hasWinCFI(), to verify this assumption.
-    HasWinCFI = true;
+  if (NeedsWinCFI) {
+    // Note that there are cases where we insert SEH opcodes in the
+    // epilogue when we had no SEH opcodes in the prologue. For
+    // example, when there is no stack frame but there are stack
+    // arguments. Insert the SEH_EpilogStart and remove it later if it
+    // we didn't emit any SEH opcodes to avoid generating WinCFI for
+    // functions that don't need it.
     BuildMI(MBB, LastPopI, DL, TII->get(AArch64::SEH_EpilogStart))
         .setMIFlag(MachineInstr::FrameDestroy);
+    EpilogStartI = LastPopI;
+    --EpilogStartI;
   }
 
   if (hasFP(MF) && AFI->hasSwiftAsyncContext()) {
@@ -2255,11 +2265,11 @@
     emitFrameOffset(
         MBB, LastPopI, DL, AArch64::SP, AArch64::FP,
         StackOffset::getFixed(-AFI->getCalleeSaveBaseToFrameRecordOffset()),
-        TII, MachineInstr::FrameDestroy, false, NeedsWinCFI);
+        TII, MachineInstr::FrameDestroy, false, NeedsWinCFI, &HasWinCFI);
   } else if (NumBytes)
     emitFrameOffset(MBB, LastPopI, DL, AArch64::SP, AArch64::SP,
                     StackOffset::getFixed(NumBytes), TII,
-                    MachineInstr::FrameDestroy, false, NeedsWinCFI);
+                    MachineInstr::FrameDestroy, false, NeedsWinCFI, &HasWinCFI);
 
   // When we are about to restore the CSRs, the CFA register is SP again.
   if (EmitCFI && hasFP(MF)) {
diff --git a/llvm/test/CodeGen/AArch64/wincfi-seh-only-in-epilogue.ll b/llvm/test/CodeGen/AArch64/wincfi-seh-only-in-epilogue.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/wincfi-seh-only-in-epilogue.ll
@@ -0,0 +1,13 @@
+; RUN: llc -mtriple=aarch64-windows %s -o - | FileCheck %s
+
+define hidden swifttailcc void @test(ptr noalias nocapture %0, ptr swiftasync %1, ptr %2, ptr noalias nocapture %3, ptr nocapture dereferenceable(8) %4, ptr %5, ptr %6, ptr %Act, ptr %Err, ptr %Res, ptr %Act.DistributedActor, ptr %Err.Error, ptr %Res.Decodable, ptr %Res.Encodable, ptr swiftself %7) #0 {
+entry:
+  ret void
+}
+
+; Check that there is no .seh_endprologue but there is seh_startepilogue/seh_endepilogue.
+; CHECK-NOT: .seh_endprologue
+; CHECK:     .seh_startepilogue
+; CHECK:     add sp, sp, #48
+; CHECK:     .seh_stackalloc 48
+; CHECK:     .seh_endepilogue