diff --git a/dlt/destinations/impl/synapse/__init__.py b/dlt/destinations/impl/synapse/__init__.py
index 53dbabc090..f6ad7369c1 100644
--- a/dlt/destinations/impl/synapse/__init__.py
+++ b/dlt/destinations/impl/synapse/__init__.py
@@ -41,6 +41,12 @@ def capabilities() -> DestinationCapabilitiesContext:
     caps.supports_transactions = True
     caps.supports_ddl_transactions = False
 
+    # Synapse throws "Some part of your SQL statement is nested too deeply. Rewrite the query or break it up into smaller queries."
+    # if number of records exceeds a certain number. Which exact number that is seems not deterministic:
+    # in tests, I've seen a query with 12230 records run succesfully on one run, but fail on a subsequent run, while the query remained exactly the same.
+    # 10.000 records is a "safe" amount that always seems to work.
+    caps.max_rows_per_insert = 10000
+
     # datetimeoffset can store 7 digits for fractional seconds
     # https://learn.microsoft.com/en-us/sql/t-sql/data-types/datetimeoffset-transact-sql?view=sql-server-ver16
     caps.timestamp_precision = 7
diff --git a/dlt/destinations/insert_job_client.py b/dlt/destinations/insert_job_client.py
index c25e8b9384..e3ce7265d1 100644
--- a/dlt/destinations/insert_job_client.py
+++ b/dlt/destinations/insert_job_client.py
@@ -38,9 +38,12 @@ def _insert(self, qualified_table_name: str, file_path: str) -> Iterator[List[st
             header = f.readline()
             writer_type = self._sql_client.capabilities.insert_values_writer_type
             if writer_type == "default":
+                sep = ","
                 # properly formatted file has a values marker at the beginning
                 values_mark = f.readline()
                 assert values_mark == "VALUES\n"
+            elif writer_type == "select_union":
+                sep = " UNION ALL"
 
             max_rows = self._sql_client.capabilities.max_rows_per_insert
 
@@ -58,10 +61,6 @@ def _insert(self, qualified_table_name: str, file_path: str) -> Iterator[List[st
                 # if there was anything left, until_nl contains the last line
                 is_eof = len(until_nl) == 0 or until_nl[-1] == ";"
                 if not is_eof:
-                    if writer_type == "default":
-                        sep = ","
-                    elif writer_type == "select_union":
-                        sep = " UNION ALL"
                     until_nl = until_nl[: -len(sep)] + ";"  # replace the separator with ";"
                 if max_rows is not None:
                     # mssql has a limit of 1000 rows per INSERT, so we need to split into separate statements
@@ -79,7 +78,7 @@ def _insert(self, qualified_table_name: str, file_path: str) -> Iterator[List[st
                             insert_sql.append("".join(chunk) + until_nl)
                         else:
                             # Replace the , with ;
-                            insert_sql.append("".join(chunk).strip()[:-1] + ";\n")
+                            insert_sql.append("".join(chunk).strip()[: -len(sep)] + ";\n")
                 else:
                     # otherwise write all content in a single INSERT INTO
                     if writer_type == "default":
diff --git a/tests/load/test_insert_job_client.py b/tests/load/test_insert_job_client.py
index 75c8440672..bd20ea9930 100644
--- a/tests/load/test_insert_job_client.py
+++ b/tests/load/test_insert_job_client.py
@@ -190,7 +190,7 @@ def test_query_split(client: InsertValuesJobClient, file_storage: FileStorage) -
             start = f"'{idx}'"
             end = ");"
         elif writer_type == "select_union":
-            start = f"ELECT '{idx}'"
+            start = f"SELECT '{idx}'"
             end = ";"
         assert fragment[-1].startswith(start)
         assert fragment[-1].endswith(end)