diff --git a/libcudacxx/include/cuda/__memory_resource/cuda_pinned_memory_resource.h b/libcudacxx/include/cuda/__memory_resource/cuda_pinned_memory_resource.h
new file mode 100644
index 00000000000..a1045e2f1cc
--- /dev/null
+++ b/libcudacxx/include/cuda/__memory_resource/cuda_pinned_memory_resource.h
@@ -0,0 +1,191 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of libcu++, the C++ Standard Library for your entire system,
+// 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
+// SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _CUDA__MEMORY_RESOURCE_CUDA_PINNED_MEMORY_RESOURCE_H
+#define _CUDA__MEMORY_RESOURCE_CUDA_PINNED_MEMORY_RESOURCE_H
+
+#include <cuda/std/detail/__config>
+
+#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
+#  pragma GCC system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
+#  pragma clang system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
+#  pragma system_header
+#endif // no system header
+
+#if !defined(_CCCL_COMPILER_MSVC_2017)
+
+#  if !defined(_CCCL_CUDA_COMPILER_NVCC) && !defined(_CCCL_CUDA_COMPILER_NVHPC)
+#    include <cuda_runtime_api.h>
+#  endif // !_CCCL_CUDA_COMPILER_NVCC && !_CCCL_CUDA_COMPILER_NVHPC
+
+#  include <cuda/__memory_resource/get_property.h>
+#  include <cuda/__memory_resource/properties.h>
+#  include <cuda/__memory_resource/resource.h>
+#  include <cuda/__memory_resource/resource_ref.h>
+#  include <cuda/std/__cuda/api_wrapper.h>
+#  include <cuda/std/detail/libcxx/include/__new/bad_alloc.h>
+
+#  if _CCCL_STD_VER >= 2014
+
+_LIBCUDACXX_BEGIN_NAMESPACE_CUDA_MR
+
+/**
+ * @brief `cuda_pinned_memory_resource` uses cudaMallocHost / cudaFreeHost for allocation/deallocation.
+ */
+class cuda_pinned_memory_resource
+{
+private:
+  unsigned int __flags_ = cudaHostAllocDefault;
+
+  static constexpr unsigned int __available_flags =
+    cudaHostAllocDefault | cudaHostAllocPortable | cudaHostAllocMapped | cudaHostAllocWriteCombined;
+
+public:
+  constexpr cuda_pinned_memory_resource(const unsigned int __flags = cudaHostAllocDefault) noexcept
+      : __flags_(__flags & __available_flags)
+  {
+    _LIBCUDACXX_ASSERT(__flags_ == __flags, "Unexpected flags passed to cuda_pinned_memory_resource");
+  }
+
+  /**
+   * @brief Allocate host memory of size at least \p __bytes.
+   * @param __bytes The size in bytes of the allocation.
+   * @param __alignment The requested alignment of the allocation.
+   * @throw cuda::cuda_error if allocation fails with a CUDA error.
+   * @return Pointer to the newly allocated memory
+   */
+  void* allocate(const size_t __bytes, const size_t __alignment = default_cuda_malloc_host_alignment) const
+  {
+    // We need to ensure that the provided alignment matches the minimal provided alignment
+    if (!__is_valid_alignment(__alignment))
+    {
+      _CUDA_VSTD_NOVERSION::__throw_bad_alloc();
+    }
+
+    void* __ptr{nullptr};
+    _CCCL_TRY_CUDA_API(::cudaMallocHost, "Failed to allocate memory with cudaMallocHost.", &__ptr, __bytes, __flags_);
+    return __ptr;
+  }
+
+  /**
+   * @brief Deallocate memory pointed to by \p __ptr.
+   * @param __ptr Pointer to be deallocated. Must have been allocated through a call to `allocate`
+   * @param __bytes The number of bytes that was passed to the `allocate` call that returned \p __ptr.
+   * @param __alignment The alignment that was passed to the `allocate` call that returned \p __ptr.
+   */
+  void deallocate(void* __ptr, const size_t, const size_t __alignment = default_cuda_malloc_host_alignment) const
+  {
+    // We need to ensure that the provided alignment matches the minimal provided alignment
+    _LIBCUDACXX_ASSERT(__is_valid_alignment(__alignment),
+                       "Invalid alignment passed to cuda_memory_resource::deallocate.");
+    _CCCL_ASSERT_CUDA_API(::cudaFreeHost, "cuda_pinned_memory_resource::deallocate failed", __ptr);
+    (void) __alignment;
+  }
+
+  /**
+   * @brief Equality comparison with another cuda_pinned_memory_resource
+   * @return Whether both cuda_pinned_memory_resource were constructed with the same flags
+   */
+  _LIBCUDACXX_NODISCARD_ATTRIBUTE constexpr bool operator==(cuda_pinned_memory_resource const& __other) const noexcept
+  {
+    return __flags_ == __other.__flags_;
+  }
+#    if _CCCL_STD_VER <= 2017
+  /**
+   * @brief Equality comparison with another cuda_pinned_memory_resource
+   * @return Whether both cuda_pinned_memory_resource were constructed with different flags
+   */
+  _LIBCUDACXX_NODISCARD_ATTRIBUTE constexpr bool operator!=(cuda_pinned_memory_resource const& __other) const noexcept
+  {
+    return __flags_ != __other.__flags_;
+  }
+#    endif // _CCCL_STD_VER <= 2017
+
+  /**
+   * @brief Equality comparison between a cuda_memory_resource and another resource
+   * @param __lhs The cuda_memory_resource
+   * @param __rhs The resource to compare to
+   * @return If the underlying types are equality comparable, returns the result of equality comparison of both
+   * resources. Otherwise, returns false.
+   */
+  template <class _Resource>
+  _LIBCUDACXX_NODISCARD_FRIEND auto operator==(cuda_pinned_memory_resource const& __lhs, _Resource const& __rhs) noexcept
+    _LIBCUDACXX_TRAILING_REQUIRES(bool)(__different_resource<cuda_pinned_memory_resource, _Resource>)
+  {
+    return resource_ref<>{const_cast<cuda_pinned_memory_resource&>(__lhs)}
+        == resource_ref<>{const_cast<_Resource&>(__rhs)};
+  }
+#    if _CCCL_STD_VER <= 2017
+  /**
+   * @copydoc cuda_pinned_memory_resource::operator<_Resource>==(cuda_pinned_memory_resource const&, _Resource const&)
+   */
+  template <class _Resource>
+  _LIBCUDACXX_NODISCARD_FRIEND auto operator==(_Resource const& __rhs, cuda_pinned_memory_resource const& __lhs) noexcept
+    _LIBCUDACXX_TRAILING_REQUIRES(bool)(__different_resource<cuda_pinned_memory_resource, _Resource>)
+  {
+    return resource_ref<>{const_cast<cuda_pinned_memory_resource&>(__lhs)}
+        == resource_ref<>{const_cast<_Resource&>(__rhs)};
+  }
+  /**
+   * @copydoc cuda_pinned_memory_resource::operator<_Resource>==(cuda_pinned_memory_resource const&, _Resource const&)
+   */
+  template <class _Resource>
+  _LIBCUDACXX_NODISCARD_FRIEND auto operator!=(cuda_pinned_memory_resource const& __lhs, _Resource const& __rhs) noexcept
+    _LIBCUDACXX_TRAILING_REQUIRES(bool)(__different_resource<cuda_pinned_memory_resource, _Resource>)
+  {
+    return resource_ref<>{const_cast<cuda_pinned_memory_resource&>(__lhs)}
+        != resource_ref<>{const_cast<_Resource&>(__rhs)};
+  }
+  /**
+   * @copydoc cuda_pinned_memory_resource::operator<_Resource>==(cuda_pinned_memory_resource const&, _Resource const&)
+   */
+  template <class _Resource>
+  _LIBCUDACXX_NODISCARD_FRIEND auto operator!=(_Resource const& __rhs, cuda_pinned_memory_resource const& __lhs) noexcept
+    _LIBCUDACXX_TRAILING_REQUIRES(bool)(__different_resource<cuda_pinned_memory_resource, _Resource>)
+  {
+    return resource_ref<>{const_cast<cuda_pinned_memory_resource&>(__lhs)}
+        != resource_ref<>{const_cast<_Resource&>(__rhs)};
+  }
+#    endif // _CCCL_STD_VER <= 2017
+
+  /**
+   * @brief Enables the `pinned_memory` property
+   */
+  friend constexpr void get_property(cuda_pinned_memory_resource const&, pinned_memory) noexcept {}
+  /**
+   * @brief Enables the `device_accessible` property
+   */
+  friend constexpr void get_property(cuda_pinned_memory_resource const&, device_accessible) noexcept {}
+  /**
+   * @brief Enables the `host_accessible` property
+   */
+  friend constexpr void get_property(cuda_pinned_memory_resource const&, host_accessible) noexcept {}
+
+  /**
+   * @brief Checks whether the passed in alignment is valid
+   */
+  static constexpr bool __is_valid_alignment(const size_t __alignment) noexcept
+  {
+    return __alignment <= default_cuda_malloc_host_alignment && (default_cuda_malloc_host_alignment % __alignment == 0);
+  }
+};
+static_assert(resource_with<cuda_pinned_memory_resource, pinned_memory>, "");
+static_assert(resource_with<cuda_pinned_memory_resource, device_accessible>, "");
+static_assert(resource_with<cuda_pinned_memory_resource, host_accessible>, "");
+
+_LIBCUDACXX_END_NAMESPACE_CUDA_MR
+
+#  endif // _CCCL_STD_VER >= 2014
+
+#endif // !_CCCL_COMPILER_MSVC_2017
+
+#endif //_CUDA__MEMORY_RESOURCE_CUDA_PINNED_MEMORY_RESOURCE_H
diff --git a/libcudacxx/include/cuda/__memory_resource/properties.h b/libcudacxx/include/cuda/__memory_resource/properties.h
index bca25a4499e..47cf321f278 100644
--- a/libcudacxx/include/cuda/__memory_resource/properties.h
+++ b/libcudacxx/include/cuda/__memory_resource/properties.h
@@ -50,6 +50,12 @@ struct host_accessible
 struct managed_memory
 {};
 
+/**
+ * @brief The \c pinned_memory property signals that the allocated memory is not pageable.
+ */
+struct pinned_memory
+{};
+
 _LIBCUDACXX_END_NAMESPACE_CUDA_MR
 
 #  endif // _CCCL_STD_VER >= 2014
diff --git a/libcudacxx/include/cuda/memory_resource b/libcudacxx/include/cuda/memory_resource
index 4a08a76d9b9..a138995aa5f 100644
--- a/libcudacxx/include/cuda/memory_resource
+++ b/libcudacxx/include/cuda/memory_resource
@@ -94,6 +94,7 @@ class resource_ref {
 
 #include <cuda/__memory_resource/cuda_managed_memory_resource.h>
 #include <cuda/__memory_resource/cuda_memory_resource.h>
+#include <cuda/__memory_resource/cuda_pinned_memory_resource.h>
 #include <cuda/__memory_resource/get_property.h>
 #include <cuda/__memory_resource/properties.h>
 #include <cuda/__memory_resource/resource.h>
diff --git a/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/allocate.pass.cpp b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/allocate.pass.cpp
new file mode 100644
index 00000000000..3141de7b016
--- /dev/null
+++ b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/allocate.pass.cpp
@@ -0,0 +1,96 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of libcu++, the C++ Standard Library for your entire system,
+// 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
+// SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11
+// UNSUPPORTED: msvc-19.16
+// UNSUPPORTED: nvrtc
+
+#include <cuda/memory_resource>
+#include <cuda/std/cassert>
+#include <cuda/std/cstdint>
+#include <cuda/stream_ref>
+
+#include "test_macros.h"
+
+void ensure_pinned_host_ptr(void* ptr)
+{
+  assert(ptr != nullptr);
+  cudaPointerAttributes attributes;
+  cudaError_t status = cudaPointerGetAttributes(&attributes, ptr);
+  assert(status == cudaSuccess);
+  assert((attributes.type == cudaMemoryTypeHost) && (attributes.devicePointer != nullptr));
+}
+
+void test(const unsigned int flag)
+{
+  cuda::mr::cuda_pinned_memory_resource res{flag};
+
+  { // allocate / deallocate
+    auto* ptr = res.allocate(42);
+    static_assert(cuda::std::is_same<decltype(ptr), void*>::value, "");
+    ensure_pinned_host_ptr(ptr);
+
+    res.deallocate(ptr, 42);
+  }
+
+  { // allocate / deallocate with alignment
+    auto* ptr = res.allocate(42, 4);
+    static_assert(cuda::std::is_same<decltype(ptr), void*>::value, "");
+    ensure_pinned_host_ptr(ptr);
+
+    res.deallocate(ptr, 42, 4);
+  }
+
+#ifndef TEST_HAS_NO_EXCEPTIONS
+  { // allocate with too small alignment
+    while (true)
+    {
+      try
+      {
+        auto* ptr = res.allocate(5, 42);
+      }
+      catch (const cuda::std::bad_alloc&)
+      {
+        break;
+      }
+      assert(false);
+    }
+  }
+
+  { // allocate with non matching alignment
+    while (true)
+    {
+      try
+      {
+        auto* ptr = res.allocate(5, 1337);
+      }
+      catch (const cuda::std::bad_alloc&)
+      {
+        break;
+      }
+      assert(false);
+    }
+  }
+#endif // TEST_HAS_NO_EXCEPTIONS
+}
+
+void test()
+{
+  test(cudaHostAllocDefault);
+  test(cudaHostAllocPortable);
+  test(cudaHostAllocMapped);
+  test(cudaHostAllocWriteCombined);
+}
+
+int main(int, char**)
+{
+  NV_IF_TARGET(NV_IS_HOST, test();)
+  return 0;
+}
diff --git a/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/equality.pass.cpp b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/equality.pass.cpp
new file mode 100644
index 00000000000..a75b0925ecd
--- /dev/null
+++ b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/equality.pass.cpp
@@ -0,0 +1,134 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of libcu++, the C++ Standard Library for your entire system,
+// 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
+// SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11
+// UNSUPPORTED: msvc-19.16
+// UNSUPPORTED: nvrtc
+
+#include <cuda/memory_resource>
+#include <cuda/std/cassert>
+#include <cuda/std/cstdint>
+#include <cuda/stream_ref>
+
+enum class AccessibilityType
+{
+  Device,
+  Host,
+};
+
+template <AccessibilityType Accessibilty>
+struct resource
+{
+  void* allocate(size_t, size_t)
+  {
+    return nullptr;
+  }
+  void deallocate(void*, size_t, size_t) {}
+
+  bool operator==(const resource&) const
+  {
+    return true;
+  }
+  bool operator!=(const resource& other) const
+  {
+    return false;
+  }
+
+  template <AccessibilityType Accessibilty2                                         = Accessibilty,
+            cuda::std::enable_if_t<Accessibilty2 == AccessibilityType::Device, int> = 0>
+  friend void get_property(const resource&, cuda::mr::pinned_memory) noexcept
+  {}
+};
+static_assert(cuda::mr::resource<resource<AccessibilityType::Host>>, "");
+static_assert(!cuda::mr::resource_with<resource<AccessibilityType::Host>, cuda::mr::pinned_memory>, "");
+static_assert(cuda::mr::resource<resource<AccessibilityType::Device>>, "");
+static_assert(cuda::mr::resource_with<resource<AccessibilityType::Device>, cuda::mr::pinned_memory>, "");
+
+template <AccessibilityType Accessibilty>
+struct async_resource : public resource<Accessibilty>
+{
+  void* allocate_async(size_t, size_t, cuda::stream_ref)
+  {
+    return nullptr;
+  }
+  void deallocate_async(void*, size_t, size_t, cuda::stream_ref) {}
+};
+static_assert(cuda::mr::async_resource<async_resource<AccessibilityType::Host>>, "");
+static_assert(!cuda::mr::async_resource_with<async_resource<AccessibilityType::Host>, cuda::mr::pinned_memory>, "");
+static_assert(cuda::mr::async_resource<async_resource<AccessibilityType::Device>>, "");
+static_assert(cuda::mr::async_resource_with<async_resource<AccessibilityType::Device>, cuda::mr::pinned_memory>, "");
+
+void test()
+{
+  cuda::mr::cuda_pinned_memory_resource first{};
+  { // comparison against a plain cuda_pinned_memory_resource
+    cuda::mr::cuda_pinned_memory_resource second{cudaHostAllocDefault};
+    assert(first == second);
+    assert(!(first != second));
+  }
+
+  { // comparison against a plain cuda_pinned_memory_resource with a different flag set
+    cuda::mr::cuda_pinned_memory_resource second{cudaHostAllocPortable};
+    assert(!(first == second));
+    assert((first != second));
+  }
+
+  { // comparison against a cuda_pinned_memory_resource wrapped inside a resource_ref<cuda::mr::pinned_memory>
+    cuda::mr::cuda_pinned_memory_resource second{};
+    cuda::mr::resource_ref<cuda::mr::pinned_memory> second_ref{second};
+    assert(first == second_ref);
+    assert(!(first != second_ref));
+    assert(second_ref == first);
+    assert(!(second_ref != first));
+  }
+
+  { // comparison against a cuda_pinned_memory_resource wrapped inside a resource_ref<>
+    cuda::mr::cuda_pinned_memory_resource second{};
+    cuda::mr::resource_ref<> second_ref{second};
+    assert(first == second_ref);
+    assert(!(first != second_ref));
+    assert(second_ref == first);
+    assert(!(second_ref != first));
+  }
+
+  { // comparison against a different resource through resource_ref
+    resource<AccessibilityType::Host> host_resource{};
+    resource<AccessibilityType::Device> device_resource{};
+    assert(!(first == host_resource));
+    assert(first != host_resource);
+    assert(!(first == device_resource));
+    assert(first != device_resource);
+
+    assert(!(host_resource == first));
+    assert(host_resource != first);
+    assert(!(device_resource == first));
+    assert(device_resource != first);
+  }
+
+  { // comparison against a different resource through resource_ref
+    async_resource<AccessibilityType::Host> host_async_resource{};
+    async_resource<AccessibilityType::Device> device_async_resource{};
+    assert(!(first == host_async_resource));
+    assert(first != host_async_resource);
+    assert(!(first == device_async_resource));
+    assert(first != device_async_resource);
+
+    assert(!(host_async_resource == first));
+    assert(host_async_resource != first);
+    assert(!(device_async_resource == first));
+    assert(device_async_resource != first);
+  }
+}
+
+int main(int, char**)
+{
+  NV_IF_TARGET(NV_IS_HOST, test();)
+  return 0;
+}
diff --git a/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/traits.pass.cpp b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/traits.pass.cpp
new file mode 100644
index 00000000000..3909ac7238e
--- /dev/null
+++ b/libcudacxx/test/libcudacxx/cuda/memory_resource/cuda_pinned_memory_resource/traits.pass.cpp
@@ -0,0 +1,31 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of libcu++, the C++ Standard Library for your entire system,
+// 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
+// SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+// UNSUPPORTED: c++03, c++11
+// UNSUPPORTED: msvc-19.16
+// UNSUPPORTED: nvrtc
+
+#include <cuda/memory_resource>
+#include <cuda/std/type_traits>
+
+using resource = cuda::mr::cuda_pinned_memory_resource;
+static_assert(!cuda::std::is_trivial<resource>::value, "");
+static_assert(!cuda::std::is_trivially_default_constructible<resource>::value, "");
+static_assert(cuda::std::is_trivially_copy_constructible<resource>::value, "");
+static_assert(cuda::std::is_trivially_move_constructible<resource>::value, "");
+static_assert(cuda::std::is_trivially_copy_assignable<resource>::value, "");
+static_assert(cuda::std::is_trivially_move_assignable<resource>::value, "");
+static_assert(cuda::std::is_trivially_destructible<resource>::value, "");
+static_assert(!cuda::std::is_empty<resource>::value, "");
+
+int main(int, char**)
+{
+  return 0;
+}