diff --git a/flang-rt/lib/runtime/CMakeLists.txt b/flang-rt/lib/runtime/CMakeLists.txt
index 332c0872e065f..5200b2b710a5e 100644
--- a/flang-rt/lib/runtime/CMakeLists.txt
+++ b/flang-rt/lib/runtime/CMakeLists.txt
@@ -21,6 +21,7 @@ set(supported_sources
   allocatable.cpp
   array-constructor.cpp
   assign.cpp
+  assign_omp.cpp
   buffer.cpp
   character.cpp
   connection.cpp
@@ -99,6 +100,7 @@ set(gpu_sources
   allocatable.cpp
   array-constructor.cpp
   assign.cpp
+  assign_omp.cpp
   buffer.cpp
   character.cpp
   connection.cpp
diff --git a/flang-rt/lib/runtime/assign_omp.cpp b/flang-rt/lib/runtime/assign_omp.cpp
new file mode 100644
index 0000000000000..10eb22a2650e1
--- /dev/null
+++ b/flang-rt/lib/runtime/assign_omp.cpp
@@ -0,0 +1,77 @@
+//===-- lib/runtime/assign_omp.cpp ----------------------------------*- C++
+//-*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "flang-rt/runtime/assign-impl.h"
+#include "flang-rt/runtime/derived.h"
+#include "flang-rt/runtime/descriptor.h"
+#include "flang-rt/runtime/stat.h"
+#include "flang-rt/runtime/terminator.h"
+#include "flang-rt/runtime/tools.h"
+#include "flang-rt/runtime/type-info.h"
+#include "flang/Runtime/assign.h"
+
+#include <omp.h>
+
+namespace Fortran::runtime {
+namespace omp {
+
+typedef int32_t OMPDeviceTy;
+
+template <typename T> static T *getDevicePtr(T *anyPtr, OMPDeviceTy ompDevice) {
+  auto voidAnyPtr = reinterpret_cast<void *>(anyPtr);
+  // If not present on the device it should already be a device ptr
+  if (!omp_target_is_present(voidAnyPtr, ompDevice))
+    return anyPtr;
+  T *device_ptr = omp_get_mapped_ptr(anyPtr, ompDevice);
+  return device_ptr;
+}
+
+RT_API_ATTRS static void Assign(Descriptor &to, const Descriptor &from,
+    Terminator &terminator, int flags, OMPDeviceTy omp_device) {
+  std::size_t toElementBytes{to.ElementBytes()};
+  std::size_t fromElementBytes{from.ElementBytes()};
+  std::size_t toElements{to.Elements()};
+  std::size_t fromElements{from.Elements()};
+
+  if (toElementBytes != fromElementBytes)
+    terminator.Crash("Assign: toElementBytes != fromElementBytes");
+  if (toElements != fromElements)
+    terminator.Crash("Assign: toElements != fromElements");
+
+  // Get base addresses and calculate length
+  void *to_base = to.raw().base_addr;
+  void *from_base = from.raw().base_addr;
+  size_t length = toElements * toElementBytes;
+
+  // Get device pointers after ensuring data is on device
+  void *to_ptr = getDevicePtr(to_base, omp_device);
+  void *from_ptr = getDevicePtr(from_base, omp_device);
+
+  // Perform copy between device pointers
+  int result = omp_target_memcpy(to_ptr, from_ptr, length,
+      /*dst_offset*/ 0, /*src_offset*/ 0, omp_device, omp_device);
+
+  if (result != 0)
+    terminator.Crash("Assign: omp_target_memcpy failed");
+  return;
+}
+
+extern "C" {
+RT_EXT_API_GROUP_BEGIN
+void RTDEF(Assign_omp)(Descriptor &to, const Descriptor &from,
+    const char *sourceFile, int sourceLine, omp::OMPDeviceTy omp_device) {
+  Terminator terminator{sourceFile, sourceLine};
+  Fortran::runtime::omp::Assign(to, from, terminator,
+      MaybeReallocate | NeedFinalization | ComponentCanBeDefinedAssignment,
+      omp_device);
+}
+
+} // extern "C"
+} // namespace omp
+} // namespace Fortran::runtime
diff --git a/flang/include/flang/Runtime/assign.h b/flang/include/flang/Runtime/assign.h
index 7d198bdcc9e89..b510d2bbfcc8b 100644
--- a/flang/include/flang/Runtime/assign.h
+++ b/flang/include/flang/Runtime/assign.h
@@ -56,6 +56,9 @@ extern "C" {
 // API for lowering assignment
 void RTDECL(Assign)(Descriptor &to, const Descriptor &from,
     const char *sourceFile = nullptr, int sourceLine = 0);
+void RTDECL(Assign_omp)(Descriptor &to, const Descriptor &from,
+    const char *sourceFile = nullptr, int sourceLine = 0,
+    int32_t omp_device = 0);
 // This variant has no finalization, defined assignment, or allocatable
 // reallocation.
 void RTDECL(AssignTemporary)(Descriptor &to, const Descriptor &from,