diff --git a/include/nanobind/stl/complex.h b/include/nanobind/stl/complex.h
new file mode 100644
index 00000000..3c4c8a1c
--- /dev/null
+++ b/include/nanobind/stl/complex.h
@@ -0,0 +1,67 @@
+/*
+    nanobind/stl/complex.h: type caster for std::complex<...>
+
+    Copyright (c) 2023 Degottex Gilles and Wenzel Jakob
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+#pragma once
+
+#include <nanobind/nanobind.h>
+#include <complex>
+
+NAMESPACE_BEGIN(NB_NAMESPACE)
+NAMESPACE_BEGIN(detail)
+
+template <typename T> struct type_caster<std::complex<T>> {
+    NB_TYPE_CASTER(std::complex<T>, const_name("complex") )
+
+    template <bool Recursive = true>
+    bool from_python(handle src, uint8_t flags,
+                     cleanup_list *cleanup) noexcept {
+        (void) flags;
+        (void) cleanup;
+
+        if (PyComplex_Check(src.ptr())) {
+            value = std::complex<T>(
+                (T) PyComplex_RealAsDouble(src.ptr()),
+                (T) PyComplex_ImagAsDouble(src.ptr())
+            );
+            return true;
+        }
+
+        if (Recursive && !PyFloat_CheckExact(src.ptr()) &&
+            !PyLong_CheckExact(src.ptr()) &&
+            PyObject_HasAttrString(src.ptr(), "imag")) {
+            try {
+                object tmp = handle(&PyComplex_Type)(src);
+                return from_python<false>(tmp, flags, cleanup);
+            } catch (...) {
+                return false;
+            }
+        }
+
+        make_caster<T> caster;
+        if (caster.from_python(src, flags, cleanup)) {
+            value = std::complex<T>(caster.operator cast_t<T>());
+            return true;
+        }
+
+        return true;
+    }
+
+    template <typename T2>
+    static handle from_cpp(T2 &&value, rv_policy policy,
+                           cleanup_list *cleanup) noexcept {
+        (void) policy;
+        (void) cleanup;
+
+        return PyComplex_FromDoubles((double) value.real(),
+                                     (double) value.imag());
+    }
+};
+
+NAMESPACE_END(detail)
+NAMESPACE_END(NB_NAMESPACE)
diff --git a/tests/test_stl.cpp b/tests/test_stl.cpp
index 6e9ca652..0ddc76f5 100644
--- a/tests/test_stl.cpp
+++ b/tests/test_stl.cpp
@@ -12,6 +12,7 @@
 #include <nanobind/stl/unordered_set.h>
 #include <nanobind/stl/set.h>
 #include <nanobind/stl/filesystem.h>
+#include <nanobind/stl/complex.h>
 
 NB_MAKE_OPAQUE(std::vector<float, std::allocator<float>>)
 
@@ -422,4 +423,20 @@ NB_MODULE(test_stl_ext, m) {
         vec.flip();
         return vec;
     });
+
+
+    m.def("complex_value_float", [](const std::complex<float>& x){
+        return x;
+    });
+    m.def("complex_value_double", [](const std::complex<double>& x){
+        return x;
+    });
+
+    m.def("complex_array_float", [](const std::vector<std::complex<float>>& x){
+        return x;
+    });
+    m.def("complex_array_double", [](const std::vector<std::complex<double>>& x){
+        return x;
+    });
+
 }
diff --git a/tests/test_stl.py b/tests/test_stl.py
index 19e42c12..bb1b1d9d 100644
--- a/tests/test_stl.py
+++ b/tests/test_stl.py
@@ -3,7 +3,6 @@
 import sys
 from common import collect, skip_on_pypy
 
-
 @pytest.fixture
 def clean():
     collect()
@@ -758,3 +757,66 @@ def test67_vector_bool():
     bool_vector = [True, False, True, False]
     result = t.flip_vector_bool(bool_vector)
     assert result == [not x for x in bool_vector]
+
+
+def test68_complex_value():
+    # double: 64bits
+    assert t.complex_value_double(1.0) == 1.0
+    assert t.complex_value_double(1.0j) == 1.0j
+    assert t.complex_value_double(0.0) == 0.0
+    assert t.complex_value_double(0.0j) == 0.0j
+    assert t.complex_value_double(0) == 0
+    assert t.complex_value_float(1.0) == 1.0
+    assert t.complex_value_float(1.0j) == 1.0j
+    assert t.complex_value_float(0.0) == 0.0
+    assert t.complex_value_float(0.0j) == 0.0j
+    assert t.complex_value_float(0) == 0
+
+    val_64 = 2.7-3.2j
+    val_32 = 2.700000047683716-3.200000047683716j
+    assert val_64 != val_32
+
+    assert t.complex_value_float(val_32) == val_32
+    assert t.complex_value_float(val_64) == val_32
+    assert t.complex_value_double(val_32) == val_32
+    assert t.complex_value_double(val_64) == val_64
+
+    try:
+        import numpy as np
+        assert t.complex_value_float(np.complex64(val_32)) == val_32
+        assert t.complex_value_float(np.complex64(val_64)) == val_32
+        assert t.complex_value_double(np.complex64(val_32)) == val_32
+        assert t.complex_value_double(np.complex64(val_64)) == val_32
+        assert t.complex_value_float(np.complex128(val_32)) == val_32
+        assert t.complex_value_float(np.complex128(val_64)) == val_32
+        assert t.complex_value_double(np.complex128(val_32)) == val_32
+        assert t.complex_value_double(np.complex128(val_64)) == val_64
+    except ImportError:
+        pass
+
+def test69_complex_array():
+    val1_64 = 2.7-3.2j
+    val1_32 = 2.700000047683716-3.200000047683716j
+    val2_64 = 3.1415
+    val2_32 = 3.1414999961853027+0j
+
+    # test 64 bit casts
+    assert t.complex_array_double([val1_64, -1j, val2_64]) == [val1_64, -0-1j, val2_64]
+
+    # test 32 bit casts
+    assert t.complex_array_float([val1_64, -1j, val2_64]) == [val1_32, (-0-1j), val2_32]
+
+    try:
+        import numpy as np
+
+        # test 64 bit casts
+        assert t.complex_array_double(np.array([val1_64, -1j, val2_64])) == [val1_64, -0-1j, val2_64]
+        assert t.complex_array_double(np.array([val1_64, -1j, val2_64],dtype=np.complex128)) == [val1_64, -0-1j, val2_64]
+        assert t.complex_array_double(np.array([val1_64, -1j, val2_64],dtype=np.complex64)) == [val1_32, -0-1j, val2_32]
+
+        # test 32 bit casts
+        assert t.complex_array_float(np.array([val1_64, -1j, val2_64])) == [val1_32, (-0-1j), val2_32]
+        assert t.complex_array_float(np.array([val1_64, -1j, val2_64],dtype=np.complex128)) == [val1_32, (-0-1j), val2_32]
+        assert t.complex_array_float(np.array([val1_64, -1j, val2_64],dtype=np.complex64)) == [val1_32, (-0-1j), val2_32]
+    except ImportError:
+        pass