diff --git a/gilda/term.py b/gilda/term.py
index 9a77bb7..04289c5 100644
--- a/gilda/term.py
+++ b/gilda/term.py
@@ -149,14 +149,31 @@ def get_identifiers_url(db, id):
         return f'https://identifiers.org/{curie}'
 
 
+def _term_key(term: Term) -> Tuple[str, str, str]:
+    return term.db, term.id, term.text
+
+
+statuses = {'curated': 1, 'name': 2, 'synonym': 3, 'former_name': 4}
+
+
+def _priority_key(term: Term) -> Tuple[int, int]:
+    """
+    Prioritize terms (that are pre-grouped by db/id/text) first
+    based on status, and if the status is the same, give priority
+    to the ones that are from primary resources
+    """
+    return (
+        statuses[term.status],
+        0 if term.db.casefold() == term.source.casefold() else 1
+    )
+
+
 def filter_out_duplicates(terms):
     logger.info('Filtering %d terms for uniqueness...' % len(terms))
-    term_key = lambda term: (term.db, term.id, term.text)
-    statuses = {'curated': 1, 'name': 2, 'synonym': 3, 'former_name': 4}
     new_terms = []
-    for _, terms in itertools.groupby(sorted(terms, key=lambda x: term_key(x)),
-                                      key=lambda x: term_key(x)):
-        terms = sorted(terms, key=lambda x: statuses[x.status])
+    for _, terms in itertools.groupby(sorted(terms, key=_term_key),
+                                      key=_term_key):
+        terms = sorted(terms, key=_priority_key)
         new_terms.append(terms[0])
     # Re-sort the terms
     new_terms = sorted(new_terms, key=lambda x: (x.text, x.db, x.id))
diff --git a/gilda/tests/test_generate_terms.py b/gilda/tests/test_generate_terms.py
index 72e541f..5b0a999 100644
--- a/gilda/tests/test_generate_terms.py
+++ b/gilda/tests/test_generate_terms.py
@@ -75,6 +75,21 @@ def test_filter_priority():
     assert term.status == 'synonym'
 
 
+def test_filter_priority_by_source():
+    term1 = Term('mekk2', 'MEKK2', 'HGNC', '6854', 'MAP3K2',
+                 'synonym', 'hgnc', '9606')
+    term2 = Term('mekk2', 'MEKK2', 'HGNC', '6854', 'MAP3K2',
+                 'synonym', 'up', '9606')
+    terms = filter_out_duplicates([term1, term2])
+    assert len(terms) == 1
+    assert terms[0] == term1
+
+    # now test the other way, to make sure order doesn't matter
+    terms = filter_out_duplicates([term2, term1])
+    assert len(terms) == 1
+    assert terms[0] == term1
+
+
 def test_get_terms_simple():
     row = {'Entry': 'P15056',
            'Gene names  (primary )': 'BRAF',