diff --git a/besskge/dataset.py b/besskge/dataset.py
index 9423ea4..f4e827a 100644
--- a/besskge/dataset.py
+++ b/besskge/dataset.py
@@ -36,6 +36,11 @@ class KGDataset:
     #: {part: int32[n_triple, {h,r,t}]}
     triples: Dict[str, NDArray[np.int32]]
 
+    #: IDs of the triples in KGDataset.triples wrt
+    #: the ordering in the original array/dataframe
+    #: from where the triples originate.
+    original_triple_ids: Dict[str, NDArray[np.int32]]
+
     #: Entity labels by ID; str[n_entity]
     entity_dict: Optional[List[str]] = None
 
@@ -90,6 +95,9 @@ def from_triples(
         and relations have already been assigned. Note that, if entities have
         types, entities of the same type need to have contiguous IDs.
         Triples are randomly split in train/validation/test sets.
+        The attribute `KGDataset.original_triple_ids` stores the IDs
+        of the triples in each split wrt the original ordering in `data`.
+
         If a pre-defined train/validation/test split is wanted, the KGDataset
         class should be instantiated manually.
 
@@ -114,22 +122,28 @@ class should be instantiated manually.
         num_valid = int(num_triples * split[1])
 
         rng = np.random.default_rng(seed=seed)
-        rng.shuffle(data, axis=0)
-
-        triples = dict()
-        triples["train"], triples["valid"], triples["test"] = np.split(
-            data, (num_train, num_train + num_valid), axis=0
+        id_shuffle = rng.permutation(np.arange(num_triples))
+        triple_ids = dict()
+        triple_ids["train"], triple_ids["valid"], triple_ids["test"] = np.split(
+            id_shuffle, (num_train, num_train + num_valid), axis=0
         )
+        triples = dict()
+        triples["train"] = data[triple_ids["train"]]
+        triples["valid"] = data[triple_ids["valid"]]
+        triples["test"] = data[triple_ids["test"]]
 
-        return cls(
+        ds = cls(
             n_entity=data[:, [0, 2]].max() + 1,
             n_relation_type=data[:, 1].max() + 1,
             entity_dict=entity_dict,
             relation_dict=relation_dict,
             type_offsets=type_offsets,
             triples=triples,
+            original_triple_ids=triple_ids,
         )
 
+        return ds
+
     @classmethod
     def from_dataframe(
         cls,
@@ -219,6 +233,9 @@ def from_dataframe(
                 relation_dict=relation_dict,
                 type_offsets=type_offsets,
                 triples=triples,
+                original_triple_ids={
+                    k: np.arange(v.shape[0]) for k, v in triples.items()
+                },
             )
 
     @classmethod
@@ -280,6 +297,7 @@ def build_ogbl_biokg(cls, root: Path) -> "KGDataset":
             relation_dict=rel_dict,
             type_offsets=type_offsets,
             triples=triples,
+            original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
             neg_heads=neg_heads,
             neg_tails=neg_tails,
         )
@@ -329,6 +347,7 @@ def build_ogbl_wikikg2(cls, root: Path) -> "KGDataset":
             relation_dict=rel_dict,
             type_offsets=None,
             triples=triples,
+            original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
             neg_heads=neg_heads,
             neg_tails=neg_tails,
         )
diff --git a/tests/test_batch_sampler.py b/tests/test_batch_sampler.py
index cfee463..91bd6ca 100644
--- a/tests/test_batch_sampler.py
+++ b/tests/test_batch_sampler.py
@@ -36,6 +36,7 @@
     relation_dict=None,
     type_offsets=None,
     triples=triples,
+    original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
     neg_heads=None,
     neg_tails=None,
 )
diff --git a/tests/test_bess.py b/tests/test_bess.py
index f0521ea..ebf7869 100644
--- a/tests/test_bess.py
+++ b/tests/test_bess.py
@@ -79,6 +79,7 @@ def test_bess_inference(
         relation_dict=None,
         type_offsets=None,
         triples=triples,
+        original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
         neg_heads=neg_heads,
         neg_tails=neg_tails,
     )
@@ -306,6 +307,7 @@ def test_bess_topk_prediction(
         relation_dict=None,
         type_offsets=None,
         triples=triples,
+        original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
         neg_heads=neg_heads,
         neg_tails=neg_tails,
     )
diff --git a/tests/test_pipeline.py b/tests/test_pipeline.py
index 4f42f23..cd54962 100644
--- a/tests/test_pipeline.py
+++ b/tests/test_pipeline.py
@@ -51,6 +51,7 @@ def test_all_scores_pipeline(
         relation_dict=None,
         type_offsets=None,
         triples=triples,
+        original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
     )
 
     partition_mode = "h_shard" if corruption_scheme == "t" else "t_shard"
diff --git a/tests/test_sharding.py b/tests/test_sharding.py
index 557d725..495e735 100644
--- a/tests/test_sharding.py
+++ b/tests/test_sharding.py
@@ -34,6 +34,7 @@
     relation_dict=None,
     type_offsets={str(i): o for i, o in enumerate(type_offsets)},
     triples=triples,
+    original_triple_ids={k: np.arange(v.shape[0]) for k, v in triples.items()},
     neg_heads=neg_heads,
     neg_tails=neg_tails,
 )