diff --git a/docs/api/mod.rst b/docs/api/mod.rst
index 6abb48b..3ead6d9 100644
--- a/docs/api/mod.rst
+++ b/docs/api/mod.rst
@@ -10,7 +10,6 @@ Most useful functions and classes are contained in submodules.
    errors
    io
    pandas
-   parquet
    record
 
 isd
diff --git a/docs/api/parquet.rst b/docs/api/parquet.rst
deleted file mode 100644
index c47be94..0000000
--- a/docs/api/parquet.rst
+++ /dev/null
@@ -1,5 +0,0 @@
-isd.parquet
-===========
-
-.. automodule:: isd.parquet
-    :members:
diff --git a/examples/examine_parquet_metadata.py b/examples/examine_parquet_metadata.py
deleted file mode 100644
index 597af0b..0000000
--- a/examples/examine_parquet_metadata.py
+++ /dev/null
@@ -1,30 +0,0 @@
-"""Given a list of one or more parquet files, examine the metadata that would be produced by writing them to parquet using this library.
-
-The data in the provided files is repartioned by a default interval.
-"""
-
-import os.path
-import sys
-from tempfile import TemporaryDirectory
-from typing import Any, List
-
-import pandas
-import pyarrow.parquet
-from pyarrow import Table
-
-import isd.io
-
-paths = sys.argv[1:]
-data_frames = [isd.io.read_to_data_frame(path) for path in paths]
-data_frame = pandas.concat(data_frames)
-data_frame = data_frame.set_index(["timestamp", "usaf_id", "ncei_id"])
-table = Table.from_pandas(data_frame)
-
-metadata_collector: List[Any] = []
-with TemporaryDirectory() as temporary_directory:
-    directory = os.path.join(temporary_directory, "isd")
-    pyarrow.parquet.write_to_dataset(
-        table, directory, metadata_collector=metadata_collector
-    )
-
-print(metadata_collector)
diff --git a/examples/read_parquet.py b/examples/read_parquet.py
deleted file mode 100644
index 1779b12..0000000
--- a/examples/read_parquet.py
+++ /dev/null
@@ -1,9 +0,0 @@
-"""Reads an ISD parquet table and prints it to stdout."""
-
-import sys
-
-import isd.parquet
-
-directory = sys.argv[1]
-data_frame = isd.parquet.read(directory).compute()
-print(data_frame)
diff --git a/setup.cfg b/setup.cfg
index 4310e25..55d639d 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -20,9 +20,7 @@ packages = find:
 python_requires = >=3.7
 install_requires =
     click ~= 8.0
-    dask[distributed] ~= 2021.08
     pandas ~= 1.3
-    pyarrow ~= 6.0
 
 
 [options.packages.find]
diff --git a/src/isd/cli.py b/src/isd/cli.py
index c256aa2..533fd04 100644
--- a/src/isd/cli.py
+++ b/src/isd/cli.py
@@ -3,15 +3,11 @@
 import dataclasses
 import itertools
 import json
-import os
-from typing import List, Optional
 
 import click
 from click import ClickException
-from dask.distributed import Client
 
 import isd.io
-import isd.parquet
 
 
 @click.group()
@@ -30,23 +26,3 @@ def record(infile: str, index: int) -> None:
             print(json.dumps(dataclasses.asdict(record), indent=4))
         else:
             raise ClickException(f"No record with index {index}")
-
-
-@main.command()
-@click.argument("INFILES", nargs=-1)
-@click.argument("DIRECTORY")
-@click.option("-c", "--client")
-def to_parquet(
-    infiles: List[str], directory: str, client: Optional[str] = None
-) -> None:
-    """Writes one or more ISD files to a parquet table."""
-    paths = []
-    for infile in infiles:
-        if os.path.isdir(infile):
-            for file_name in os.listdir(infile):
-                paths.append(os.path.abspath(os.path.join(infile, file_name)))
-        else:
-            paths.append(os.path.abspath(infile))
-    if client:
-        Client(client)
-    isd.parquet.write(paths, directory)
diff --git a/src/isd/parquet.py b/src/isd/parquet.py
deleted file mode 100644
index d42f0f3..0000000
--- a/src/isd/parquet.py
+++ /dev/null
@@ -1,40 +0,0 @@
-from typing import List
-
-import dask.dataframe
-import dask.delayed
-
-import isd.io
-
-DEFAULT_REPARTITION_FREQUENCY = "2w"
-
-
-def write(
-    paths: List[str],
-    directory: str,
-    append: bool = False,
-    repartition_frequency: str = DEFAULT_REPARTITION_FREQUENCY,
-) -> None:
-    """Writes the ISD data located in `paths` to a partitioned parquet table at `directory`.
-
-    Uses dask under the hood to read the data and write the output file.
-    """
-    if append:
-        data_frame = read(directory)
-        last_division = data_frame.divisions[-1]
-        data_frames = [
-            dask.delayed(isd.io.read_to_data_frame)(path, since=last_division)
-            for path in paths
-        ]
-    else:
-        data_frames = [dask.delayed(isd.io.read_to_data_frame)(path) for path in paths]
-    data_frame = dask.dataframe.from_delayed(data_frames)
-    data_frame = data_frame.persist()
-    data_frame = data_frame.set_index("timestamp").repartition(
-        freq=repartition_frequency
-    )
-    data_frame.to_parquet(directory, append=append)
-
-
-def read(directory: str) -> dask.dataframe.DataFrame:
-    """Reads a parquet table as a dask data frame."""
-    return dask.dataframe.read_parquet(directory)
diff --git a/tests/test_parquet.py b/tests/test_parquet.py
deleted file mode 100644
index bbf0c11..0000000
--- a/tests/test_parquet.py
+++ /dev/null
@@ -1,21 +0,0 @@
-import os.path
-from tempfile import TemporaryDirectory
-from typing import List
-
-import isd.parquet
-
-
-def test_write_then_read(paths: List[str]) -> None:
-    with TemporaryDirectory() as temporary_directory:
-        directory = os.path.join(temporary_directory, "isd")
-        isd.parquet.write(paths, directory)
-        isd.parquet.read(directory).compute()
-
-
-def test_append(half_path: str, uncompressed_path: str) -> None:
-    with TemporaryDirectory() as temporary_directory:
-        directory = os.path.join(temporary_directory, "isd")
-        isd.parquet.write([half_path], directory)
-        isd.parquet.write([uncompressed_path], directory, append=True)
-        data_frame = isd.parquet.read(directory).compute()
-    assert len(data_frame) == 500