diff --git a/lisa/regression.py b/lisa/regression.py
index 6f9687dec49aa047591cde8563415bb47cbc054e..2ec91f686914a8614ef19adc1d4d468ac1e7fc68 100644
--- a/lisa/regression.py
+++ b/lisa/regression.py
@@ -190,7 +190,9 @@ def compute_regressions(old_list, new_list, remove_tags=[], **kwargs):
     :param old_list: old series of :class:`exekall.engine.FrozenVal`
     :type old_list: list(exekall.engine.FrozenVal)
 
-    :param new_list: new series of :class:`exekall.engine.FrozenVal`
+    :param new_list: new series of :class:`exekall.engine.FrozenVal`. Values
+        with a UUID that is also present in `old_list` will be removed from
+        that list before the regressions are computed.
     :type new_list: list(exekall.engine.FrozenVal)
 
     :param remove_tags: remove the given list of tags from the IDs before
@@ -200,6 +202,27 @@ def compute_regressions(old_list, new_list, remove_tags=[], **kwargs):
 
     :param kwargs: extra :meth:`RegressionResult.from_result_list` parameters
     """
+
+    def get_uuids(froz_val_list):
+        return {
+            froz_val.uuid
+            for froz_val in froz_val_list
+        }
+
+    def dedup_list(froz_val_list):
+        common_uuid = get_uuids(old_list) & get_uuids(new_list)
+        return [
+            froz_val
+            for froz_val in froz_val_list
+            if froz_val.uuid not in common_uuid
+        ]
+
+    # Remove from the new_list all the FrozenVal that were carried from the
+    # old_list sequence. That is important since running "exekall run --load-db"
+    # will contain both new and old data, so old data needs to be filtered out
+    # before we can actually compare the two sets.
+    new_list = dedup_list(new_list)
+
     def get_id(froz_val):
         id_ = froz_val.get_id(qual=False, with_tags=True)