diff --git a/pymatgen/analysis/local_env.py b/pymatgen/analysis/local_env.py
index 6ec084015ca..aaa2b8336fc 100644
--- a/pymatgen/analysis/local_env.py
+++ b/pymatgen/analysis/local_env.py
@@ -3794,6 +3794,10 @@ class CrystalNN(NearNeighbors):
     algorithm can also modify probability using smooth distance cutoffs as well as Pauling
     electronegativity differences. The output can either be the most probable coordination
     environment or a weighted list of coordination environments.
+    Please note that the default weights have been benchmarked for inorganic crystal structures.
+    For MOFs or molecular crystals, weights and cutoffs likely will need to be adapted.
+    A starting point could be:
+    CrystalNN(x_diff_weight = 1.5, search_cutoff = 4.5)
     """
 
     NNData = namedtuple("NNData", ["all_nninfo", "cn_weights", "cn_nninfo"])