diff --git a/python/ray/tune/examples/mlflow_example.py b/python/ray/tune/examples/mlflow_example.py
index 68be24139..368726c0b 100644
--- a/python/ray/tune/examples/mlflow_example.py
+++ b/python/ray/tune/examples/mlflow_example.py
@@ -41,7 +41,9 @@ if __name__ == "__main__":
         num_samples=5,
         loggers=DEFAULT_LOGGERS + (MLFLowLogger, ),
         config={
-            "mlflow_experiment_id": experiment_id,
+            "logger_config": {
+                "mlflow_experiment_id": experiment_id,
+            },
             "width": tune.sample_from(
                 lambda spec: 10 + int(90 * random.random())),
             "height": tune.sample_from(lambda spec: int(100 * random.random()))
diff --git a/python/ray/tune/logger.py b/python/ray/tune/logger.py
index d2fae3723..a1edf9cbd 100644
--- a/python/ray/tune/logger.py
+++ b/python/ray/tune/logger.py
@@ -78,9 +78,12 @@ class MLFLowLogger(Logger):
     """
 
     def _init(self):
+        logger_config = self.config.get("logger_config", {})
         from mlflow.tracking import MlflowClient
-        client = MlflowClient()
-        run = client.create_run(self.config.get("mlflow_experiment_id"))
+        client = MlflowClient(
+            tracking_uri=logger_config.get("mlflow_tracking_uri"),
+            registry_uri=logger_config.get("mlflow_registry_uri"))
+        run = client.create_run(logger_config.get("mlflow_experiment_id"))
         self._run_id = run.info.run_id
         for key, value in self.config.items():
             client.log_param(self._run_id, key, value)
diff --git a/rllib/agents/trainer.py b/rllib/agents/trainer.py
index b2a3c1972..b21c2fff7 100644
--- a/rllib/agents/trainer.py
+++ b/rllib/agents/trainer.py
@@ -370,6 +370,10 @@ COMMON_CONFIG: TrainerConfigDict = {
         "replay_mode": "independent",
     },
 
+    # === Logger ===
+    # Define logger-specific configuration to be used inside Logger
+    "logger_config": {},
+
     # === Replay Settings ===
     # The number of contiguous environment steps to replay at once. This may
     # be set to greater than 1 to support recurrent models.