diff --git a/tests/post_training/README.md b/tests/post_training/README.md
index af2639c8ec7..975637fc79d 100644
--- a/tests/post_training/README.md
+++ b/tests/post_training/README.md
@@ -122,3 +122,26 @@ Run test with calibration dataset having batch-size=10 for all models:
 ```bash
 pytest --data=<path_to_datasets> --batch-size 10 tests/post_training/test_quantize_conformance.py
 ```
+
+## Reference data
+
+The reference data section outlines the expected format for defining reference values used during parallel testing.
+
+```yml
+<Name from model scopes>_backend_<BACKEND>:
+  metric_value: <expected value>
+```
+
+> [!IMPORTANT]
+> The reference file is used for parallel testing.
+> The path to the *_reference_data.yaml files is used during testing and should not be changed without updating Jenkins scripts.
+
+### Marking tests as xfail
+
+To mark a test as expected to fail (xfail) when a validation metric does not meet expectations, add the following line to the reference data:
+
+```yml
+<Name from model scopes>_backend_<BACKEND>:
+  ...
+  metrics_xfail_reason: "Issue-<jira ticket number>"
+```
diff --git a/tests/post_training/data/wc_reference_data.yaml b/tests/post_training/data/wc_reference_data.yaml
index 4083aab523f..5874e4c507e 100644
--- a/tests/post_training/data/wc_reference_data.yaml
+++ b/tests/post_training/data/wc_reference_data.yaml
@@ -26,9 +26,9 @@ tinyllama_data_aware_gptq_backend_OV:
   metric_value: 0.87134
   num_int4: 94
   num_int8: 124
-  atol: 0.0004  # issue 148819
+  metrics_xfail_reason: "Issue-148819"
 tinyllama_scale_estimation_per_channel_backend_OV:
   metric_value: 0.81389
   num_int4: 188
   num_int8: 124
-  atol: 0.006  # issue 148819
+  metrics_xfail_reason: "Issue-148819"
diff --git a/tests/post_training/pipelines/base.py b/tests/post_training/pipelines/base.py
index 519668a491a..426870c9e97 100644
--- a/tests/post_training/pipelines/base.py
+++ b/tests/post_training/pipelines/base.py
@@ -36,6 +36,7 @@
 from tools.memory_monitor import memory_monitor_context
 
 DEFAULT_VAL_THREADS = 4
+METRICS_XFAIL_REASON = "metrics_xfail_reason"
 
 
 class BackendType(Enum):
@@ -307,6 +308,7 @@ def validate(self) -> None:
         if metric_value is not None and metric_value_fp32 is not None:
             self.run_info.metric_diff = round(self.run_info.metric_value - self.reference_data["metric_value_fp32"], 5)
 
+        status_msg = None
         if (
             metric_value is not None
             and metric_reference is not None
@@ -314,9 +316,13 @@ def validate(self) -> None:
         ):
             if metric_value < metric_reference:
                 status_msg = f"Regression: Metric value is less than reference {metric_value} < {metric_reference}"
-                raise ValueError(status_msg)
             if metric_value > metric_reference:
                 status_msg = f"Improvement: Metric value is better than reference {metric_value} > {metric_reference}"
+
+        if status_msg is not None:
+            if METRICS_XFAIL_REASON in self.reference_data:
+                self.run_info.status = f"XFAIL: {self.reference_data[METRICS_XFAIL_REASON]} - {status_msg}"
+            else:
                 raise ValueError(status_msg)
 
     def run(self) -> None:
diff --git a/tests/post_training/test_quantize_conformance.py b/tests/post_training/test_quantize_conformance.py
index de3b0bc96d1..34addb8a2b5 100644
--- a/tests/post_training/test_quantize_conformance.py
+++ b/tests/post_training/test_quantize_conformance.py
@@ -366,3 +366,5 @@ def test_weight_compression(
 
     if err_msg:
         pytest.fail(err_msg)
+    if run_info.status is not None and run_info.status.startswith("XFAIL:"):
+        pytest.xfail(run_info.status)