diff --git a/scripts/load_test_data.sh b/scripts/load_test_data.sh
deleted file mode 100755
index a6e975f8..00000000
--- a/scripts/load_test_data.sh
+++ /dev/null
@@ -1,41 +0,0 @@
-# Copyright (c) 2017 The PyBigQuery Authors
-#
-# Permission is hereby granted, free of charge, to any person obtaining a copy of
-# this software and associated documentation files (the "Software"), to deal in
-# the Software without restriction, including without limitation the rights to
-# use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
-# the Software, and to permit persons to whom the Software is furnished to do so,
-# subject to the following conditions:
-#
-# The above copyright notice and this permission notice shall be included in all
-# copies or substantial portions of the Software.
-#
-# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
-# FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
-# COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
-# IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
-# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-bq mk test_pybigquery
-bq mk --data_location=asia-northeast1 test_pybigquery_location
-bq mk test_pybigquery_alt
-
-bq rm -f -t test_pybigquery.sample
-bq rm -f -t test_pybigquery_alt.sample_alt
-bq rm -f -t test_pybigquery.sample_one_row
-bq rm -f -t test_pybigquery.sample_dml
-bq rm -f -t test_pybigquery.sample_view
-bq rm -f -t test_pybigquery_location.sample_one_row
-
-bq mk --table --schema=$(dirname $0)/schema.json --time_partitioning_field timestamp --clustering_fields integer,string --description 'A sample table containing most data types.' test_pybigquery.sample
-bq mk --table --schema=$(dirname $0)/schema.json --time_partitioning_field timestamp --clustering_fields integer,string test_pybigquery_alt.sample_alt
-bq load --source_format=NEWLINE_DELIMITED_JSON --schema=$(dirname $0)/schema.json test_pybigquery.sample $(dirname $0)/sample.json
-
-bq mk --table --schema=$(dirname $0)/schema.json --time_partitioning_field timestamp --clustering_fields integer,string test_pybigquery.sample_one_row
-bq load --source_format=NEWLINE_DELIMITED_JSON --schema=$(dirname $0)/schema.json test_pybigquery.sample_one_row $(dirname $0)/sample_one_row.json
-
-bq --location=asia-northeast1 load --source_format=NEWLINE_DELIMITED_JSON --schema=$(dirname $0)/schema.json test_pybigquery_location.sample_one_row $(dirname $0)/sample_one_row.json
-bq mk --schema=$(dirname $0)/schema.json -t test_pybigquery.sample_dml
-
-bq mk --use_legacy_sql=false --view 'SELECT string FROM test_pybigquery.sample' test_pybigquery.sample_view
diff --git a/tests/system/conftest.py b/tests/system/conftest.py
new file mode 100644
index 00000000..5f694133
--- /dev/null
+++ b/tests/system/conftest.py
@@ -0,0 +1,111 @@
+# Copyright 2021 The PyBigQuery Authors
+#
+# Use of this source code is governed by an MIT-style
+# license that can be found in the LICENSE file or at
+# https://opensource.org/licenses/MIT.
+
+import pathlib
+
+import pytest
+from google.cloud import bigquery
+
+from typing import List
+
+
+DATA_DIR = pathlib.Path(__file__).parent / "data"
+
+
+def load_sample_data(
+    full_table_id: str,
+    bigquery_client: bigquery.Client,
+    bigquery_schema: List[bigquery.SchemaField],
+    filename: str = "sample.json",
+):
+    # Delete the table first. Even though we can use WRITE_TRUNCATE, the load
+    # job fails if properties such as table description do not match.
+    bigquery_client.delete_table(full_table_id, not_found_ok=True)
+    sample_config = bigquery.LoadJobConfig()
+    sample_config.destination_table_description = (
+        "A sample table containing most data types."
+    )
+    sample_config.schema = bigquery_schema
+    sample_config.time_partitioning = bigquery.TimePartitioning(field="timestamp")
+    sample_config.clustering_fields = ["integer", "string"]
+    sample_config.source_format = bigquery.SourceFormat.NEWLINE_DELIMITED_JSON
+    sample_config.write_disposition = bigquery.WriteDisposition.WRITE_TRUNCATE
+
+    with open(DATA_DIR / filename, "rb") as data_file:
+        return bigquery_client.load_table_from_file(
+            data_file, full_table_id, job_config=sample_config,
+        )
+
+
+@pytest.fixture(scope="session")
+def bigquery_client():
+    return bigquery.Client()
+
+
+@pytest.fixture(scope="session")
+def bigquery_schema(bigquery_client: bigquery.Client):
+    return bigquery_client.schema_from_json(DATA_DIR / "schema.json")
+
+
+@pytest.fixture(scope="session", autouse=True)
+def bigquery_dataset(
+    bigquery_client: bigquery.Client, bigquery_schema: List[bigquery.SchemaField]
+):
+    project_id = bigquery_client.project
+    dataset_id = "test_pybigquery"
+    dataset = bigquery.Dataset(f"{project_id}.{dataset_id}")
+    dataset = bigquery_client.create_dataset(dataset, exists_ok=True)
+    empty_table = bigquery.Table(
+        f"{project_id}.{dataset_id}.sample_dml", schema=bigquery_schema
+    )
+    view = bigquery.Table(f"{project_id}.{dataset_id}.sample_view",)
+    view.view_query = f"SELECT string FROM `{dataset_id}.sample`"
+    job1 = load_sample_data(
+        f"{project_id}.{dataset_id}.sample", bigquery_client, bigquery_schema
+    )
+    job2 = load_sample_data(
+        f"{project_id}.{dataset_id}.sample_one_row",
+        bigquery_client,
+        bigquery_schema,
+        filename="sample_one_row.json",
+    )
+    bigquery_client.create_table(empty_table, exists_ok=True)
+    job1.result()
+    job2.result()
+    bigquery_client.create_table(view, exists_ok=True)
+    return dataset_id
+
+
+@pytest.fixture(scope="session", autouse=True)
+def bigquery_alt_dataset(
+    bigquery_client: bigquery.Client, bigquery_schema: List[bigquery.SchemaField]
+):
+    project_id = bigquery_client.project
+    dataset_id = "test_pybigquery_alt"
+    dataset = bigquery.Dataset(f"{project_id}.{dataset_id}")
+    dataset = bigquery_client.create_dataset(dataset, exists_ok=True)
+    job = load_sample_data(
+        f"{project_id}.{dataset_id}.sample_alt", bigquery_client, bigquery_schema
+    )
+    job.result()
+    return dataset_id
+
+
+@pytest.fixture(scope="session", autouse=True)
+def bigquery_regional_dataset(bigquery_client, bigquery_schema):
+    project_id = bigquery_client.project
+    dataset_id = "test_pybigquery_location"
+    dataset = bigquery.Dataset(f"{project_id}.{dataset_id}")
+    dataset.location = "asia-northeast1"
+    dataset = bigquery_client.create_dataset(dataset, exists_ok=True)
+    job = load_sample_data(
+        f"{project_id}.{dataset_id}.sample_one_row",
+        bigquery_client,
+        bigquery_schema,
+        filename="sample_one_row.json",
+    )
+    job.result()
+    return dataset_id
diff --git a/scripts/sample.json b/tests/system/data/sample.json
similarity index 100%
rename from scripts/sample.json
rename to tests/system/data/sample.json
diff --git a/scripts/sample_one_row.json b/tests/system/data/sample_one_row.json
similarity index 100%
rename from scripts/sample_one_row.json
rename to tests/system/data/sample_one_row.json
diff --git a/scripts/schema.json b/tests/system/data/schema.json
similarity index 100%
rename from scripts/schema.json
rename to tests/system/data/schema.json