diff --git a/bimmer_connected/api/authentication.py b/bimmer_connected/api/authentication.py
index b300e8a7..ceca4790 100644
--- a/bimmer_connected/api/authentication.py
+++ b/bimmer_connected/api/authentication.py
@@ -84,30 +84,34 @@ async def async_auth_flow(self, request: httpx.Request) -> AsyncGenerator[httpx.
 
         # Try getting a response
         response: httpx.Response = (yield request)
-        await response.aread()
-        prev_response_code: int = 0
 
-        # Retry 3 times on 401 or 429
-        for _ in range(3):
-            # Handle "classic" 401 Unauthorized and try getting a new token
-            # We don't want to call the auth endpoint too many times, so we only do it once per 401
-            if response.status_code == 401 and response.status_code != prev_response_code:
-                prev_response_code = response.status_code
-                async with self.login_lock:
-                    _LOGGER.debug("Received unauthorized response, refreshing token.")
-                    await self.login()
-                request.headers["authorization"] = f"Bearer {self.access_token}"
-                request.headers["bmw-session-id"] = self.session_id
-                response = yield request
+        # return directly if first response was successful
+        if response.is_success:
+            return
 
+        await response.aread()
+
+        # First check against 429 Too Many Requests and 403 Quota Exceeded
+        retry_count = 0
+        while (
+            response.status_code == 429 or (response.status_code == 403 and "quota" in response.text.lower())
+        ) and retry_count < 3:
             # Quota errors can either be 429 Too Many Requests or 403 Quota Exceeded (instead of 403 Forbidden)
-            elif response.status_code == 429 or (response.status_code == 403 and "quota" in response.text.lower()):
-                prev_response_code = response.status_code
-                await response.aread()
-                wait_time = get_retry_wait_time(response)
-                _LOGGER.debug("Sleeping %s seconds due to 429 Too Many Requests", wait_time)
-                await asyncio.sleep(wait_time)
-                response = yield request
+            wait_time = get_retry_wait_time(response)
+            _LOGGER.debug("Sleeping %s seconds due to 429 Too Many Requests", wait_time)
+            await asyncio.sleep(wait_time)
+            response = yield request
+            await response.aread()
+            retry_count += 1
+
+        # Handle 401 Unauthorized and try getting a new token
+        if response.status_code == 401:
+            async with self.login_lock:
+                _LOGGER.debug("Received unauthorized response, refreshing token.")
+                await self.login()
+            request.headers["authorization"] = f"Bearer {self.access_token}"
+            request.headers["bmw-session-id"] = self.session_id
+            response = yield request
 
         # Raise if request still was not successful
         try:
diff --git a/bimmer_connected/tests/test_account.py b/bimmer_connected/tests/test_account.py
index 759961ff..609d6166 100644
--- a/bimmer_connected/tests/test_account.py
+++ b/bimmer_connected/tests/test_account.py
@@ -508,6 +508,9 @@ async def test_429_retry_with_login_raise_vehicles(bmw_fixture: respx.Router):
             httpx.Response(401),
             httpx.Response(429, json=json_429),
             httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
         ]
     )
 
@@ -533,6 +536,52 @@ async def test_multiple_401(bmw_fixture: respx.Router):
             await account.get_vehicles()
 
 
+@pytest.mark.asyncio
+async def test_401_after_429_ok(bmw_fixture: respx.Router):
+    """Test the error handling, when a 401 is received after exactly 3 429."""
+    account = MyBMWAccount(TEST_USERNAME, TEST_PASSWORD, TEST_REGION)
+    await account.get_vehicles()
+
+    json_429 = {"statusCode": 429, "message": "Rate limit is exceeded. Try again in 2 seconds."}
+
+    # Recover after 3 429 and 1 401
+    bmw_fixture.get("/eadrax-vcs/v4/vehicles").mock(
+        side_effect=[
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(401),
+            *[httpx.Response(200, json={})] * 100,  # Just simulate OK responses from now on
+        ]
+    )
+    with mock.patch("asyncio.sleep", new_callable=mock.AsyncMock):
+        await account.get_vehicles()
+    assert len(account.vehicles) == get_fingerprint_state_count()
+
+
+@pytest.mark.asyncio
+async def test_401_after_429_fail(bmw_fixture: respx.Router):
+    """Test the error handling, when a 401 is received after exactly 3 429."""
+    account = MyBMWAccount(TEST_USERNAME, TEST_PASSWORD, TEST_REGION)
+
+    json_429 = {"statusCode": 429, "message": "Rate limit is exceeded. Try again in 2 seconds."}
+
+    # Fail after 3 429 and 1 401 with another 429
+    bmw_fixture.get("/eadrax-vcs/v4/vehicles").mock(
+        side_effect=[
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(429, json=json_429),
+            httpx.Response(401),
+            httpx.Response(429, json=json_429),
+        ]
+    )
+
+    with mock.patch("asyncio.sleep", new_callable=mock.AsyncMock):
+        with pytest.raises(MyBMWQuotaError):
+            await account.get_vehicles()
+
+
 @pytest.mark.asyncio
 async def test_403_quota_exceeded_vehicles_usa(caplog, bmw_fixture: respx.Router):
     """Test 403 quota issues for vehicle state and fail if it happens too often."""