diff --git a/include/mimalloc.h b/include/mimalloc.h
index 57f31336..d43a4419 100644
--- a/include/mimalloc.h
+++ b/include/mimalloc.h
@@ -456,6 +456,124 @@ template<class T> struct mi_stl_allocator {
 
 template<class T1,class T2> bool operator==(const mi_stl_allocator<T1>& , const mi_stl_allocator<T2>& ) mi_attr_noexcept { return true; }
 template<class T1,class T2> bool operator!=(const mi_stl_allocator<T1>& , const mi_stl_allocator<T2>& ) mi_attr_noexcept { return false; }
+
+#if (__cplusplus >= 201103L) || (_MSC_VER > 1900)  // C++11
+#include <memory>      // std::shared_ptr
+
+// STL allocator allocation in a specific heap
+template<class T> struct mi_heap_stl_allocator {
+  typedef T                 value_type;
+  typedef std::size_t       size_type;
+  typedef std::ptrdiff_t    difference_type;
+  typedef value_type& reference;
+  typedef value_type const& const_reference;
+  typedef value_type* pointer;
+  typedef value_type const* const_pointer;
+  template <class U> struct rebind { typedef mi_heap_stl_allocator<U> other; };
+
+  mi_heap_stl_allocator() { 
+    mi_heap_t* hp = mi_heap_new();
+    this->heap.reset(hp, heap_delete);
+  }
+  mi_heap_stl_allocator(mi_heap_t* hp) : heap(hp) { }    /* will not delete or destroy the passed in heap */
+  mi_heap_stl_allocator(const mi_heap_stl_allocator& x) mi_attr_noexcept : heap(x.heap) { }
+  template<class U> mi_heap_stl_allocator(const mi_heap_stl_allocator<U>& x) mi_attr_noexcept : heap(x.heap) { }  
+  mi_heap_stl_allocator  select_on_container_copy_construction() const { return *this; }
+  void deallocate(T* p, size_type) { mi_free(p); }
+
+  #if (__cplusplus >= 201703L)  // C++17
+  mi_decl_nodiscard T* allocate(size_type count) { return static_cast<T*>(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(T))); }
+  mi_decl_nodiscard T* allocate(size_type count, const void*) { return allocate(count); }
+  #else
+  mi_decl_nodiscard pointer allocate(size_type count, const void* = 0) { return static_cast<pointer>(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(value_type))); }
+  #endif
+
+  #if ((__cplusplus >= 201103L) || (_MSC_VER > 1900))  // C++11
+  using propagate_on_container_copy_assignment = std::true_type;
+  using propagate_on_container_move_assignment = std::true_type;
+  using propagate_on_container_swap = std::true_type;
+  using is_always_equal = std::false_type;
+  template <class U, class ...Args> void construct(U* p, Args&& ...args) { ::new(p) U(std::forward<Args>(args)...); }
+  template <class U> void destroy(U* p) mi_attr_noexcept { p->~U(); }
+  #else
+  void construct(pointer p, value_type const& val) { ::new(p) value_type(val); }
+  void destroy(pointer p) { p->~value_type(); }
+  #endif
+
+  size_type     max_size() const mi_attr_noexcept { return (PTRDIFF_MAX / sizeof(value_type)); }
+  pointer       address(reference x) const { return &x; }
+  const_pointer address(const_reference x) const { return &x; }
+
+  void collect(bool force) { mi_heap_collect(this->heap.get(), force); }
+
+protected:
+  std::shared_ptr<mi_heap_t> heap;
+
+private:
+  static void heap_delete(mi_heap_t* hp)  { if (hp != NULL) { mi_heap_delete(hp); } }
+};
+
+template<class T1, class T2> bool operator==(const mi_heap_stl_allocator<T1>& x, const mi_heap_stl_allocator<T2>& y) mi_attr_noexcept { return (x.heap == y.heap); }
+template<class T1, class T2> bool operator!=(const mi_heap_stl_allocator<T1>& x, const mi_heap_stl_allocator<T2>& y) mi_attr_noexcept { return (x.heap != y.heap); }
+
+
+// STL allocator allocation in a specific heap, where `free` does nothing and 
+// the heap is destroyed in one go on destruction -- use with care!
+template<class T> struct mi_heap_destroy_stl_allocator {
+  typedef T                 value_type;
+  typedef std::size_t       size_type;
+  typedef std::ptrdiff_t    difference_type;
+  typedef value_type& reference;
+  typedef value_type const& const_reference;
+  typedef value_type* pointer;
+  typedef value_type const* const_pointer;
+  template <class U> struct rebind { typedef mi_heap_destroy_stl_allocator<U> other; };
+
+  mi_heap_destroy_stl_allocator() {
+    mi_heap_t* hp = mi_heap_new();
+    this->heap.reset(hp, heap_destroy);
+  }
+  mi_heap_destroy_stl_allocator(mi_heap_t* hp) : heap(hp) { }    /* will not delete or destroy the passed-in heap; nor free any allocated objects it allocates in the heap! */
+  mi_heap_destroy_stl_allocator(const mi_heap_destroy_stl_allocator& x) mi_attr_noexcept : heap(x.heap) { }
+  template<class U> mi_heap_destroy_stl_allocator(const mi_heap_destroy_stl_allocator<U>& x) mi_attr_noexcept : heap(x.heap) { }
+  mi_heap_destroy_stl_allocator  select_on_container_copy_construction() const { return *this; }
+  void deallocate(T* p, size_type) { /* do nothing as we destroy the heap on destruct. */ }
+
+  #if (__cplusplus >= 201703L)  // C++17
+  mi_decl_nodiscard T* allocate(size_type count) { return static_cast<T*>(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(T))); }
+  mi_decl_nodiscard T* allocate(size_type count, const void*) { return allocate(count); }
+  #else
+  mi_decl_nodiscard pointer allocate(size_type count, const void* = 0) { return static_cast<pointer>(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(value_type))); }
+  #endif
+
+  #if ((__cplusplus >= 201103L) || (_MSC_VER > 1900))  // C++11
+  using propagate_on_container_copy_assignment = std::true_type;
+  using propagate_on_container_move_assignment = std::true_type;
+  using propagate_on_container_swap = std::true_type;
+  using is_always_equal = std::false_type;
+  template <class U, class ...Args> void construct(U* p, Args&& ...args) { ::new(p) U(std::forward<Args>(args)...); }
+  template <class U> void destroy(U* p) mi_attr_noexcept { p->~U(); }
+  #else
+  void construct(pointer p, value_type const& val) { ::new(p) value_type(val); }
+  void destroy(pointer p) { p->~value_type(); }
+  #endif
+
+  size_type     max_size() const mi_attr_noexcept { return (PTRDIFF_MAX / sizeof(value_type)); }
+  pointer       address(reference x) const { return &x; }
+  const_pointer address(const_reference x) const { return &x; }
+
+// protected:
+  std::shared_ptr<mi_heap_t> heap;
+
+private:
+  static void heap_destroy(mi_heap_t* hp) { if (hp != NULL) { mi_heap_destroy(hp); } }
+};
+
+template<class T1, class T2> bool operator==(const mi_heap_destroy_stl_allocator<T1>& x, const mi_heap_destroy_stl_allocator<T2>& y) mi_attr_noexcept { return (x.heap == y.heap); }
+template<class T1, class T2> bool operator!=(const mi_heap_destroy_stl_allocator<T1>& x, const mi_heap_destroy_stl_allocator<T2>& y) mi_attr_noexcept { return (x.heap != y.heap); }
+
+#endif // C++11
+
 #endif // __cplusplus
 
 #endif
diff --git a/src/alloc.c b/src/alloc.c
index 84a9fb43..21ad3bb3 100644
--- a/src/alloc.c
+++ b/src/alloc.c
@@ -91,7 +91,10 @@ extern inline void* _mi_page_malloc(mi_heap_t* heap, mi_page_t* page, size_t siz
 
 static inline mi_decl_restrict void* mi_heap_malloc_small_zero(mi_heap_t* heap, size_t size, bool zero) mi_attr_noexcept {
   mi_assert(heap != NULL);
-  mi_assert(heap->thread_id == 0 || heap->thread_id == _mi_thread_id()); // heaps are thread local
+  #if MI_DEBUG 
+  const uintptr_t tid = _mi_thread_id();
+  mi_assert(heap->thread_id == 0 || heap->thread_id == tid); // heaps are thread local
+  #endif
   mi_assert(size <= MI_SMALL_SIZE_MAX);
 #if (MI_PADDING)
   if (size == 0) {
diff --git a/test/main-override.cpp b/test/main-override.cpp
index f748c75a..f5cb3668 100644
--- a/test/main-override.cpp
+++ b/test/main-override.cpp
@@ -36,6 +36,8 @@ static void fail_aslr();              // issue #372
 static void tsan_numa_test();         // issue #414
 static void strdup_test();     // issue #445
 
+static void test_stl_allocators();
+
 int main() {
   mi_stats_reset();  // ignore earlier allocations
   heap_thread_free_large();
@@ -46,6 +48,8 @@ int main() {
   tsan_numa_test();
   strdup_test();
 
+  test_stl_allocators();
+
   test_mt_shutdown();
   //fail_aslr();
   mi_stats_print(NULL);
@@ -122,6 +126,43 @@ static bool test_stl_allocator2() {
   return vec.size() == 0;
 }
 
+static bool test_stl_allocator3() {
+  std::vector<int, mi_heap_stl_allocator<int> > vec;
+  vec.push_back(1);
+  vec.pop_back();
+  return vec.size() == 0;
+}
+
+static bool test_stl_allocator4() {
+  std::vector<some_struct, mi_heap_stl_allocator<some_struct> > vec;
+  vec.push_back(some_struct());
+  vec.pop_back();
+  return vec.size() == 0;
+}
+
+static bool test_stl_allocator5() {
+  std::vector<int, mi_heap_destroy_stl_allocator<int> > vec;
+  vec.push_back(1);
+  vec.pop_back();
+  return vec.size() == 0;
+}
+
+static bool test_stl_allocator6() {
+  std::vector<some_struct, mi_heap_destroy_stl_allocator<some_struct> > vec;
+  vec.push_back(some_struct());
+  vec.pop_back();
+  return vec.size() == 0;
+}
+
+static void test_stl_allocators() {
+  test_stl_allocator1();
+  test_stl_allocator2();
+  test_stl_allocator3();
+  test_stl_allocator4();
+  test_stl_allocator5();
+  test_stl_allocator6();
+}
+
 // issue 445
 static void strdup_test() {
 #ifdef _MSC_VER