diff --git a/postgres_copy/copy_to.py b/postgres_copy/copy_to.py
index 3b32a11..2b198b3 100644
--- a/postgres_copy/copy_to.py
+++ b/postgres_copy/copy_to.py
@@ -10,6 +10,7 @@
 from psycopg2.extensions import adapt
 from django.db.models.sql.query import Query
 from django.db.models.sql.compiler import SQLCompiler
+
 logger = logging.getLogger(__name__)
 
 
@@ -17,6 +18,7 @@ class SQLCopyToCompiler(SQLCompiler):
     """
     Custom SQL compiler for creating a COPY TO query (postgres backend only).
     """
+
     def setup_query(self):
         """
         Extend the default SQLCompiler.setup_query to add re-ordering of items in select.
@@ -34,7 +36,7 @@ def setup_query(self):
                 )
                 self.select.append(selection)
 
-    def execute_sql(self, csv_path_or_obj=None):
+    def execute_sql(self, csv_path_or_obj=None, client_encoding=None):
         """
         Run the COPY TO query.
         """
@@ -46,6 +48,15 @@ def execute_sql(self, csv_path_or_obj=None):
 
         # use stdout to avoid file permission issues
         with connections[self.using].cursor() as c:
+            # set client encoding to adapted params
+            if client_encoding is None:
+                client_encoding = c.connection.encoding
+            elif client_encoding != c.connection.encoding:
+                raise ValueError('client_encoding does not match'
+                                 ' db encoding: {} != {}'.format(client_encoding, c.connection.encoding))
+            for p in adapted_params:
+                if hasattr(p, 'encoding'):
+                    p.encoding = client_encoding if client_encoding else p.encoding
             # compile the SELECT query
             select_sql = self.as_sql()[0] % adapted_params
             # then the COPY TO query
@@ -86,6 +97,7 @@ class CopyToQuery(Query):
     """
     Represents a "copy to" SQL query.
     """
+
     def get_compiler(self, using=None, connection=None):
         """
         Return a SQLCopyToCompiler object.
diff --git a/postgres_copy/managers.py b/postgres_copy/managers.py
index 6df7244..fb188db 100644
--- a/postgres_copy/managers.py
+++ b/postgres_copy/managers.py
@@ -215,9 +215,13 @@ def to_csv(self, csv_path=None, *fields, **kwargs):
         escape_char = kwargs.get('escape', None)
         query.copy_to_escape = "ESCAPE '{}'".format(escape_char) if escape_char else ""
 
+        # Client encoding
+        client_encoding = kwargs.get('client_encoding', None)
+
         # Run the query
         compiler = query.get_compiler(self.db, connection=connection)
-        data = compiler.execute_sql(csv_path)
+
+        data = compiler.execute_sql(csv_path, client_encoding=client_encoding)
 
         # If no csv_path is provided, then the query will come back as a string.
         if csv_path is None:
diff --git a/tests/data/special_names.csv b/tests/data/special_names.csv
new file mode 100644
index 0000000..03ff4e0
--- /dev/null
+++ b/tests/data/special_names.csv
@@ -0,0 +1,5 @@
+NAME,NUMBER,DATE
+ben,1,2012-01-01
+joe,2,2012-01-02
+jane,3,2012-01-03
+björn,4,2012-01-04
diff --git a/tests/tests.py b/tests/tests.py
index 4db9563..895993d 100644
--- a/tests/tests.py
+++ b/tests/tests.py
@@ -39,6 +39,7 @@ def setUp(self):
             'matching_headers.csv'
         )
         self.secondarydb_path = os.path.join(self.data_dir, 'secondary_db.csv')
+        self.special_names_path = os.path.join(self.data_dir, 'special_names.csv')
 
     def tearDown(self):
         MockObject.objects.all().delete()
@@ -227,6 +228,32 @@ def test_filter(self, _):
             [i['name'] for i in reader]
         )
 
+    @mock.patch("django.db.connection.validate_no_atomic_block")
+    def test_filter_special_names(self, _):
+        self._load_objects(self.special_names_path)
+        MockObject.objects.filter(name="björn").to_csv(self.export_path)
+        reader = csv.DictReader(open(self.export_path, 'r'))
+        self.assertTrue(
+            ['BJÖRN'],
+            [i['name'] for i in reader]
+        )
+
+    @mock.patch("django.db.connection.validate_no_atomic_block")
+    def test_filter_special_names_encoding_error(self, _):
+        with self.assertRaises(ValueError):
+            self._load_objects(self.special_names_path)
+            MockObject.objects.filter(name="björn").to_csv(self.export_path, client_encoding='latin1')
+
+    @mock.patch("django.db.connection.validate_no_atomic_block")
+    def test_filter_number(self, _):
+        # test filter by number (int), because adapted int parameters do not have an encoding attribute.
+        self._load_objects(self.name_path)
+        MockObject.objects.filter(number=3).to_csv(self.export_path)
+        reader = csv.DictReader(open(self.export_path, 'r'))
+        self.assertTrue(
+            ['JANE'],
+            [i['name'] for i in reader])
+
     @mock.patch("django.db.connection.validate_no_atomic_block")
     def test_fewer_fields(self, _):
         self._load_objects(self.name_path)