diff --git a/mlte/spec/model.py b/mlte/spec/model.py
index 193cfd09..425f371b 100644
--- a/mlte/spec/model.py
+++ b/mlte/spec/model.py
@@ -4,6 +4,7 @@
 Model implementation for the Spec artifact.
 """
 
+import json
 from typing import Any, Dict, List, Literal, Optional
 
 from mlte.artifact.type import ArtifactType
@@ -25,6 +26,18 @@ class ConditionModel(BaseModel):
     value_class: str
     """A string indicating the full module and class name of the Value used to generate this condition."""
 
+    def args_to_json_str(self) -> str:
+        """
+        Serialize the model arguments field into a string.
+        :return: The JSON str representation of the model
+        """
+        # First convert whole thing, to see if arguments will trigger error (and if so, just let it bubble up).
+        self.to_json()
+
+        # Now convert only the actual arguments.
+        json_args = json.dumps(self.arguments)
+        return json_args
+
 
 class PropertyModel(BaseModel):
     """A description of a property."""
diff --git a/mlte/store/artifact/underlying/rdbs/factory_spec.py b/mlte/store/artifact/underlying/rdbs/factory_spec.py
index f5b6e23a..64d9c458 100644
--- a/mlte/store/artifact/underlying/rdbs/factory_spec.py
+++ b/mlte/store/artifact/underlying/rdbs/factory_spec.py
@@ -47,7 +47,7 @@ def create_spec_db_from_model(
             condition_obj = DBCondition(
                 name=condition.name,
                 measurement_id=measurement_id,
-                arguments=json.dumps(condition.arguments),
+                arguments=condition.args_to_json_str(),
                 callback=condition.callback,
                 value_class=condition.value_class,
                 property=property_obj,
diff --git a/test/spec/test_condition.py b/test/spec/test_condition.py
index e38d5eb9..deb23c39 100644
--- a/test/spec/test_condition.py
+++ b/test/spec/test_condition.py
@@ -6,7 +6,13 @@
 
 from __future__ import annotations
 
+from typing import Any
+
+import pytest
+
+from mlte._private import serializing
 from mlte.evidence.metadata import EvidenceMetadata, Identifier
+from mlte.model.serialization_error import SerializationError
 from mlte.spec.condition import Condition
 from mlte.spec.model import ConditionModel
 from mlte.validation.result import Failure, Success
@@ -16,6 +22,8 @@
 class TestValue:
     """Test value class to test build_condition method."""
 
+    data: Any
+
     @classmethod
     def in_between(cls, arg1: float, arg2: float) -> Condition:
         """Checks if the value is in between the arguments."""
@@ -30,6 +38,20 @@ def in_between(cls, arg1: float, arg2: float) -> Condition:
         )
         return condition
 
+    @classmethod
+    def in_between_complex(cls, arg1: float, arg2: TestValue) -> Condition:
+        """Checks if the value is in between the arguments."""
+        condition: Condition = Condition.build_condition(
+            lambda real: Success(
+                f"Real magnitude {real.value} between {arg1} and {arg2}"
+            )
+            if real.value > arg1 and real.value < arg2
+            else Failure(
+                f"Real magnitude {real.value} not between {arg1} and {arg2}"
+            )
+        )
+        return condition
+
 
 def test_condition_model() -> None:
     """A Condition model can be serialized and deserialized."""
@@ -43,7 +65,7 @@ def test_condition_model() -> None:
         ConditionModel(
             name="greater_than",
             arguments=[1, 2],
-            callback=Condition.encode_callback(
+            callback=serializing.encode_callable(
                 lambda real: Success("Real magnitude 2 less than threshold 3")
                 if 3 < 4
                 else Failure("Real magnitude 2 exceeds threshold 1")
@@ -87,3 +109,11 @@ def test_call_condition():
 
     result = condition(Real(ev, 11.0))
     assert str(result) == "Failure"
+
+
+def test_non_serializable_argument():
+    test_value = TestValue()
+    condition = test_value.in_between_complex(1.0, test_value)
+
+    with pytest.raises(SerializationError):
+        _ = condition.to_model().to_json()