diff --git a/perceval/backends/mozilla/kitsune.py b/perceval/backends/mozilla/kitsune.py
index 8cd8500..8221333 100644
--- a/perceval/backends/mozilla/kitsune.py
+++ b/perceval/backends/mozilla/kitsune.py
@@ -278,6 +278,7 @@ def get_question_answers(self, question_id):
         """Retrieve all answers for a question from older to newer (updated)"""
 
         page = KitsuneClient.FIRST_PAGE
+        failures = 0
 
         while True:
             api_answers_url = urijoin(self.base_url, '/answer') + '/'
@@ -286,9 +287,21 @@ def get_question_answers(self, question_id):
                 "question": question_id,
                 "ordering": "updated"
             }
-
-            answers_raw = self.fetch(api_answers_url, params)
-            yield answers_raw
+            try:
+                answers_raw = self.fetch(api_answers_url, params)
+                yield answers_raw
+                failures = 0
+            except requests.exceptions.HTTPError as e:
+                failures += 1
+                if e.response.status_code == 500:
+                    if failures > self.max_retries:
+                        logger.error(f"Problem getting Kitsune answers for question id {question_id}. "
+                                     f"Skipping answers for this question.")
+                        return  # Stop generator
+                    time.sleep(2 ** failures)
+                    continue
+                else:
+                    raise
 
             answers = json.loads(answers_raw)
             if not answers['next']:
@@ -323,6 +336,9 @@ def fetch(self, url, params):
                     self.sleep_for_rate_limit()
                 else:
                     raise ex
+            finally:
+                # Avoids to do requests too fast
+                time.sleep(0.5)
 
         raise HttpClientError(cause="Max retries exceeded")
 
diff --git a/releases/unreleased/handle-kitsune-server-errors.yml b/releases/unreleased/handle-kitsune-server-errors.yml
new file mode 100644
index 0000000..4c9a610
--- /dev/null
+++ b/releases/unreleased/handle-kitsune-server-errors.yml
@@ -0,0 +1,9 @@
+---
+title: Handle Kitsune server errors
+category: fixed
+author: Jose Javier Merchante <jjmerchante@bitergia.com>
+issue: null
+notes: >
+  This change adds retry logic for HTTP 500 errors when
+  fetching answers. Ignores the answers after several
+  retries.
diff --git a/tests/test_kitsune.py b/tests/test_kitsune.py
index c553d26..f2719b5 100644
--- a/tests/test_kitsune.py
+++ b/tests/test_kitsune.py
@@ -25,8 +25,10 @@
 import json
 import time
 import unittest
+import unittest.mock
 
 import httpretty
+import requests
 
 from grimoirelab_toolkit.datetime import str_to_datetime
 
@@ -42,6 +44,7 @@
 KITSUNE_API = KITSUNE_SERVER_URL + '/api/2/'
 KITSUNE_API_QUESTION = KITSUNE_SERVER_URL + '/api/2/question/'
 KITSUNE_API_ANSWER = KITSUNE_SERVER_URL + '/api/2/answer/'
+KITSUNE_API_ANSWERS_ERROR = KITSUNE_SERVER_URL + '/api/2/answer/?question=12345&page=1&ordering=updated'
 
 KITSUNE_SERVER_FAIL_DATE = '3000-01-01'
 KITSUNE_SERVER_RATELIMIT_DATE = '1980-01-01'
@@ -132,6 +135,11 @@ def request_callback(method, uri, headers):
                                    httpretty.Response(body=request_callback)
                                ])
 
+        httpretty.register_uri(httpretty.GET,
+                               KITSUNE_API_ANSWERS_ERROR,
+                               match_querystring=True,
+                               status=500)
+
 
 class TestKitsuneBackend(unittest.TestCase):
     """Kitsune backend tests"""
@@ -459,6 +467,25 @@ def test_rate_limit_error(self):
         }
         self.assertDictEqual(req.querystring, expected)
 
+    def test_get_question_answers_retries_and_stops_on_500(self):
+        """Test get question answers stops after max_retries on repeated HTTP 500"""
+
+        client = KitsuneClient(KITSUNE_SERVER_URL, max_retries=2)
+        question_id = 12345
+
+        # Simulate HTTP 500 error
+        error = requests.exceptions.HTTPError()
+        error.response = unittest.mock.Mock(status_code=500)
+
+        with unittest.mock.patch.object(client, 'fetch', side_effect=error) as mock_fetch, \
+             unittest.mock.patch('time.sleep') as mock_sleep, \
+             self.assertLogs(level='ERROR') as cm:
+            result = list(client.get_question_answers(question_id))
+            self.assertEqual(result, [])
+            self.assertIn(f"Problem getting Kitsune answers for question id {question_id}", cm.output[-1])
+            self.assertEqual(mock_fetch.call_count, 3)  # initial + 2 retries
+            self.assertEqual(mock_sleep.call_count, 2)  # sleep called for each retry
+
 
 if __name__ == "__main__":
     unittest.main(warnings='ignore')