diff --git a/lib/crewai/src/crewai/flow/expressions.py b/lib/crewai/src/crewai/flow/expressions.py
index 5fd9eb522..2250124df 100644
--- a/lib/crewai/src/crewai/flow/expressions.py
+++ b/lib/crewai/src/crewai/flow/expressions.py
@@ -21,6 +21,25 @@ _CEL_MACROS_WITH_LOCAL_BINDINGS = frozenset(
 )
 
 
+def _find_cel_eval_error(value: Any) -> Exception | None:
+    from celpy.evaluation import CELEvalError
+
+    if isinstance(value, CELEvalError):
+        return value
+    if isinstance(value, dict):
+        for key, item in value.items():
+            if (error := _find_cel_eval_error(key)) is not None:
+                return error
+            if (error := _find_cel_eval_error(item)) is not None:
+                return error
+        return None
+    if isinstance(value, (list, tuple)):
+        for item in value:
+            if (error := _find_cel_eval_error(item)) is not None:
+                return error
+    return None
+
+
 def _stringify_cel_value(value: Any) -> str:
     from celpy.adapter import CELJSONEncoder
 
@@ -336,6 +355,8 @@ class Expression:
                 Expression._compile_cel(expression, environment=environment)
             )
             result = program.evaluate(cast(Context, json_to_cel(context)))
+            if (eval_error := _find_cel_eval_error(result)) is not None:
+                raise eval_error
             return json.loads(json.dumps(result, cls=CELJSONEncoder))
         except Exception as e:
             raise ExpressionError(
diff --git a/lib/crewai/tests/test_flow_from_definition.py b/lib/crewai/tests/test_flow_from_definition.py
index c35893256..f5ac6ee15 100644
--- a/lib/crewai/tests/test_flow_from_definition.py
+++ b/lib/crewai/tests/test_flow_from_definition.py
@@ -2952,6 +2952,52 @@ def test_expression_template_empty_context_overrides_stored_context():
         expression.render_template({})
 
 
+@pytest.mark.parametrize(
+    "expression",
+    [
+        "{'a': 1/0}",
+        "{'a': 1, 'b': state.missing}",
+        "{'a': {'b': 1/0}}",
+        "{'a': [1/0]}",
+    ],
+)
+def test_expression_raises_for_cel_eval_error_returned_as_data(expression):
+    """celpy returns a map literal holding a CELEvalError instead of raising it."""
+    from crewai.flow.expressions import Expression, ExpressionError
+
+    with pytest.raises(ExpressionError, match="failed to evaluate CEL expression"):
+        Expression(expression, context={"state": {"score": 90}}).evaluate()
+
+
+def test_expression_nested_cel_eval_error_reports_underlying_cause():
+    from crewai.flow.expressions import Expression, ExpressionError
+
+    expression = Expression("{'a': 1/0}", context={"state": {}})
+
+    with pytest.raises(ExpressionError, match="modulus or divide by zero"):
+        expression.evaluate()
+
+
+def test_expression_keeps_short_circuited_cel_errors():
+    """Errors that CEL logic intentionally silences must still evaluate."""
+    from crewai.flow.expressions import Expression
+
+    context = {"state": {"tags": ["a", "b"]}}
+
+    assert Expression("{'ok': false && 1/0 == 1}", context=context).evaluate() == {
+        "ok": False
+    }
+    assert Expression("{'ok': true || 1/0 == 1}", context=context).evaluate() == {
+        "ok": True
+    }
+    assert (
+        Expression(
+            "state.tags.exists(t, t == 'a' || 1/0 == 1)", context=context
+        ).evaluate()
+        is True
+    )
+
+
 def test_expression_action_can_route_like_if_else():
     yaml_str = f"""
 schema: crewai.flow/v1