diff --git a/fafbseg/flywire/synapses.py b/fafbseg/flywire/synapses.py
index 6ecc5dc..538b821 100644
--- a/fafbseg/flywire/synapses.py
+++ b/fafbseg/flywire/synapses.py
@@ -805,14 +805,24 @@ def get_adjacency(
         _check_ids(both, materialization=materialization, dataset=dataset)
 
     columns = ["pre_pt_root_id", "post_pt_root_id", "cleft_score", "id"]
+    sv_cols = ["pre_pt_supervoxel_id", "post_pt_supervoxel_id"]
 
-    if materialization == "live":
+    if materialization == "live" and filtered:
+        raise ValueError(
+            "It is currently not possible to fetch filtered "
+            "synapses in live queries. You can set `filtered=False` "
+            "but please be aware that this will query the "
+            "unfiltered synapse table. See docs for details."
+        )
+    elif materialization == "live":
         func = partial(
             retry(client.materialize.live_query),
             table=client.materialize.synapse_table,
             timestamp=dt.datetime.utcnow(),
-            select_columns=columns,
-        )
+            # nb there is a bug in CAVE which causes empty results if we don't
+            # ask for supervoxels
+            select_columns=columns + sv_cols,
+        )        
     elif filtered:
         has_view = "valid_connection_v2" in client.materialize.get_views(
             materialization
@@ -858,7 +868,7 @@ def get_adjacency(
         for k in range(0, len(targets), batch_size):
             target_batch = targets[k : k + batch_size]
 
-            if not filtered:
+            if not filtered or materialization == "live":
                 filter_in_dict = dict(
                     post_pt_root_id=target_batch, pre_pt_root_id=source_batch
                 )
@@ -868,15 +878,16 @@ def get_adjacency(
                         post_pt_root_id=target_batch, pre_pt_root_id=source_batch
                     )
                 )
-
-            this = func(filter_in_dict=filter_in_dict)
+            this = func(filter_in_dict=filter_in_dict)            
 
             # We need to drop the .attrs (which contain meta data from queries)
             # Otherwise we run into issues when concatenating
             this.attrs = {}
 
             if not this.empty:
-                syn.append(this)
+                syn.append(this.drop(
+                    sv_cols, axis=1, errors="ignore"
+                ))
 
     # Combine results from batches
     if len(syn):
@@ -1106,6 +1117,7 @@ def get_connectivity(
         _check_ids(ids, materialization=materialization, dataset=dataset)
 
     columns = ["pre_pt_root_id", "post_pt_root_id", "cleft_score", "id"]
+    sv_cols = ["pre_pt_supervoxel_id", "post_pt_supervoxel_id"]
 
     if transmitters:
         columns += ["gaba", "ach", "glut", "oct", "ser", "da"]
@@ -1122,7 +1134,7 @@ def get_connectivity(
             retry(client.materialize.live_query),
             table=client.materialize.synapse_table,
             timestamp=dt.datetime.utcnow(),
-            select_columns=columns,
+            select_columns=columns + sv_cols,
         )
     elif filtered:
         has_view = "valid_connection_v2" in client.materialize.get_views(
@@ -1169,20 +1181,23 @@ def get_connectivity(
     ):
         batch = ids[i : i + batch_size]
         if upstream:
-            if not filtered:
+            if not filtered or materialization == "live":
                 filter_in_dict = dict(post_pt_root_id=batch)
             else:
                 filter_in_dict = dict(synapses_nt_v1=dict(post_pt_root_id=batch))
             syn.append(func(filter_in_dict=filter_in_dict))
         if downstream:
-            if not filtered:
+            if not filtered or materialization == "live":
                 filter_in_dict = dict(pre_pt_root_id=batch)
             else:
                 filter_in_dict = dict(synapses_nt_v1=dict(pre_pt_root_id=batch))
             syn.append(func(filter_in_dict=filter_in_dict))
 
-    # Drop attrs to avoid issues when concatenating
+    # Some clean-up
     for df in syn:
+        # Drop supervoxel columns (if they exist)
+        df.drop(sv_cols, axis=1, errors="ignore", inplace=True)
+        # Drop `attrs`` to avoid issues when concatenating
         df.attrs = {}
 
     # Combine results from batches