diff --git a/tests/test_api.py b/tests/test_api.py
index 1acc26f..251331a 100644
--- a/tests/test_api.py
+++ b/tests/test_api.py
@@ -242,6 +242,33 @@ def test_pathlib():
         assert test_str == toml.dumps(o, encoder=toml.TomlPathlibEncoder())
 
 
+def test_dataclass():
+    if (3, 7) <= sys.version_info:
+        from dataclasses import dataclass, asdict
+
+        @dataclass
+        class TestDataClassIn():
+            a: int
+
+        @dataclass
+        class TestDataClass():
+            a: int
+            c: str
+            nested: TestDataClassIn
+            nested_arr: list
+
+        dc = TestDataClass(
+                a=1, c="ccc",
+                nested=TestDataClassIn(a=1),
+                nested_arr=[TestDataClassIn(a=10),
+                            TestDataClassIn(a=100)]
+             )
+        o = {"dc_on_root": dc, "dcarr_on_root": [TestDataClassIn(a=-1), TestDataClassIn(a=-2)]}
+        o_expected = {"dc_on_root": asdict(dc), "dcarr_on_root": [asdict(v) for v in o["dcarr_on_root"]]}
+
+        assert o_expected == toml.loads(toml.dumps(o, encoder=toml.TomlDataclassEncoder()))
+
+
 def test_comment_preserve_decoder_encoder():
     test_str = """[[products]]
 name = "Nail"
diff --git a/toml/__init__.py b/toml/__init__.py
index 7719ac2..3614eac 100644
--- a/toml/__init__.py
+++ b/toml/__init__.py
@@ -23,3 +23,4 @@
 TomlNumpyEncoder = encoder.TomlNumpyEncoder
 TomlPreserveCommentEncoder = encoder.TomlPreserveCommentEncoder
 TomlPathlibEncoder = encoder.TomlPathlibEncoder
+TomlDataclassEncoder = encoder.TomlDataclassEncoder
diff --git a/toml/encoder.py b/toml/encoder.py
index bf17a72..b4b2e59 100644
--- a/toml/encoder.py
+++ b/toml/encoder.py
@@ -179,6 +179,9 @@ def dump_value(self, v):
         # Evaluate function (if it exists) else return v
         return dump_fn(v) if dump_fn is not None else self.dump_funcs[str](v)
 
+    def _preprocess_section(self, contents):
+        return contents
+
     def dump_sections(self, o, sup):
         retstr = ""
         if sup != "" and sup[-1] != ".":
@@ -188,16 +191,17 @@ def dump_sections(self, o, sup):
         for section in o:
             section = unicode(section)
             qsection = section
+            contents = self._preprocess_section(o[section])
             if not re.match(r'^[A-Za-z0-9_-]+$', section):
                 qsection = _dump_str(section)
-            if not isinstance(o[section], dict):
+            if not isinstance(contents, dict):
                 arrayoftables = False
-                if isinstance(o[section], list):
-                    for a in o[section]:
+                if isinstance(contents, list):
+                    for a in contents:
                         if isinstance(a, dict):
                             arrayoftables = True
                 if arrayoftables:
-                    for a in o[section]:
+                    for a in contents:
                         arraytabstr = "\n"
                         arraystr += "[[" + sup + qsection + "]]\n"
                         s, d = self.dump_sections(a, sup + qsection)
@@ -221,14 +225,14 @@ def dump_sections(self, o, sup):
                             d = newd
                         arraystr += arraytabstr
                 else:
-                    if o[section] is not None:
+                    if contents is not None:
                         retstr += (qsection + " = " +
-                                   unicode(self.dump_value(o[section])) + '\n')
-            elif self.preserve and isinstance(o[section], InlineTableDict):
+                                   unicode(self.dump_value(contents)) + '\n')
+            elif self.preserve and isinstance(contents, InlineTableDict):
                 retstr += (qsection + " = " +
-                           self.dump_inline_table(o[section]))
+                           self.dump_inline_table(contents))
             else:
-                retdict[qsection] = o[section]
+                retdict[qsection] = contents
         retstr += arraystr
         return (retstr, retdict)
 
@@ -302,3 +306,19 @@ def dump_value(self, v):
             if isinstance(v, pathlib.PurePath):
                 v = str(v)
         return super(TomlPathlibEncoder, self).dump_value(v)
+
+class TomlDataclassEncoder(TomlEncoder):
+
+    def _preprocess_section(self, contents):
+        if (3, 7) <= sys.version_info:
+            import dataclasses
+            if dataclasses.is_dataclass(contents):
+                contents = dataclasses.asdict(contents)
+            elif isinstance(contents, list):
+                contents, _contents = [], contents
+                for c in _contents:
+                    if dataclasses.is_dataclass(c):
+                        c = dataclasses.asdict(c)
+                    contents.append(c)
+
+        return super(TomlDataclassEncoder, self)._preprocess_section(contents)
\ No newline at end of file
diff --git a/toml/encoder.pyi b/toml/encoder.pyi
index 194a358..584234c 100644
--- a/toml/encoder.pyi
+++ b/toml/encoder.pyi
@@ -32,3 +32,6 @@ class TomlPreserveCommentEncoder(TomlEncoder):
 
 class TomlPathlibEncoder(TomlEncoder):
     def dump_value(self, v: Any): ...
+
+class TomlDataclassEncoder(TomlEncoder):
+    def dump_value(self, v: Any): ...