diff --git a/by-dataframe/pandas/read_pandas.py b/by-dataframe/pandas/read_pandas.py
new file mode 100644
index 00000000..0b7b9ce9
--- /dev/null
+++ b/by-dataframe/pandas/read_pandas.py
@@ -0,0 +1,99 @@
+"""
+About
+=====
+
+Evaluate reading data from CrateDB into pandas dataframes.
+
+Example program to demonstrate reading data in batches from CrateDB into
+pandas, using SQLAlchemy, supporting urllib3 vs. psycopg3.
+
+
+Setup
+=====
+::
+
+    pip install --upgrade click pandas 'sqlalchemy-cratedb[all]'
+
+
+Synopsis
+========
+::
+
+    # Run CrateDB.
+    docker run --rm -it --publish=4200:4200 --publish=5432:5432 crate:latest
+
+    # Use CrateDB, either talking HTTP, or PostgreSQL wire protocol.
+    python read_pandas.py --dburi='crate+urllib3://crate@localhost:4200'
+    python read_pandas.py --dburi='crate+psycopg://crate@localhost:5432'
+
+    # Use bulk size parameter to exercise paging.
+    python read_pandas.py --bulk-size 50
+
+    # Use CrateDB Cloud.
+    python read_pandas.py --dburi='crate://admin:<PASSWORD>@example.aks1.westeurope.azure.cratedb.net:4200?ssl=true'
+
+
+Details
+=======
+To watch the HTTP traffic to your local CrateDB instance, invoke::
+
+    sudo ngrep -d lo0 -Wbyline port 4200
+
+"""
+import logging
+
+import click
+import pandas as pd
+import sqlalchemy as sa
+from pueblo.util.logging import setup_logging
+
+logger = logging.getLogger(__name__)
+
+
+SQLALCHEMY_LOGGING = True
+
+
+class DatabaseWorkload:
+
+    table_name = "testdrive_pandas"
+
+    def __init__(self, dburi: str):
+        self.dburi = dburi
+
+    def get_engine(self, **kwargs):
+        return sa.create_engine(self.dburi, **kwargs)
+
+    def process(self, bulk_size: int):
+        """
+        Exercise different insert methods of pandas, SQLAlchemy, and CrateDB.
+        """
+
+        logger.info(f"Connecting to {self.dburi}")
+        logger.info(f"Reading data with bulk_size={bulk_size}")
+
+        engine = self.get_engine()
+        frames = pd.read_sql(sql="SELECT * FROM sys.summits;", con=engine, chunksize=bulk_size)
+        for df in frames:
+            print(df)
+
+
+def tweak_log_levels(level=logging.INFO):
+
+    # Enable SQLAlchemy logging.
+    if SQLALCHEMY_LOGGING:
+        logging.getLogger("sqlalchemy").setLevel(level)
+
+
+@click.command()
+@click.option("--dburi", type=str, default="crate://localhost:4200", required=False, help="SQLAlchemy database connection URI.")
+@click.option("--bulk-size", type=int, default=5_000, required=False, help="Bulk size / chunk size.")
+@click.help_option()
+def main(dburi: str, bulk_size: int):
+    setup_logging()
+    tweak_log_levels()
+    dbw = DatabaseWorkload(dburi=dburi)
+    dbw.process(bulk_size)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/by-dataframe/pandas/requirements.txt b/by-dataframe/pandas/requirements.txt
index 280ed991..90449e2b 100644
--- a/by-dataframe/pandas/requirements.txt
+++ b/by-dataframe/pandas/requirements.txt
@@ -2,4 +2,4 @@ click<9
 colorlog<7
 pandas==2.2.*
 pueblo==0.0.9
-sqlalchemy-cratedb==0.39.0
+sqlalchemy-cratedb[all] @ git+https://github.com/crate-workbench/sqlalchemy-cratedb@amo/postgresql-async
diff --git a/by-dataframe/pandas/test.py b/by-dataframe/pandas/test_insert.py
similarity index 100%
rename from by-dataframe/pandas/test.py
rename to by-dataframe/pandas/test_insert.py
diff --git a/by-dataframe/pandas/test_read.py b/by-dataframe/pandas/test_read.py
new file mode 100644
index 00000000..99e90d82
--- /dev/null
+++ b/by-dataframe/pandas/test_read.py
@@ -0,0 +1,16 @@
+import shlex
+import subprocess
+
+
+def run(command: str):
+    subprocess.check_call(shlex.split(command))
+
+
+def test_read_urllib3():
+    cmd = "time python read_pandas.py --dburi=crate+urllib3://crate@localhost:4200"
+    run(cmd)
+
+
+def test_read_psycopg3():
+    cmd = "time python read_pandas.py --dburi=crate+psycopg://crate@localhost:5432"
+    run(cmd)