diff --git a/python/doc/source/development.rst b/python/doc/source/development.rst
index b5aba6c53ef..55b3efdad17 100644
--- a/python/doc/source/development.rst
+++ b/python/doc/source/development.rst
@@ -84,7 +84,7 @@ from conda-forge:
    conda create -y -q -n pyarrow-dev \
          python=3.6 numpy six setuptools cython pandas pytest \
          cmake flatbuffers rapidjson boost-cpp thrift-cpp snappy zlib \
-         brotli jemalloc -c conda-forge
+         brotli jemalloc lz4-c zstd -c conda-forge
    source activate pyarrow-dev
 
 
diff --git a/python/pyarrow/_parquet.pyx b/python/pyarrow/_parquet.pyx
index bbe52033526..6081a64fd90 100644
--- a/python/pyarrow/_parquet.pyx
+++ b/python/pyarrow/_parquet.pyx
@@ -629,7 +629,10 @@ cdef class ParquetWriter:
         cdef CTable* ctable = table.table
 
         if row_group_size is None or row_group_size == -1:
-            row_group_size = ctable.num_rows()
+            if ctable.num_rows() > 0:
+                row_group_size = ctable.num_rows()
+            else:
+                row_group_size = 1
         elif row_group_size == 0:
             raise ValueError('Row group size cannot be 0')
 
diff --git a/python/pyarrow/tests/test_parquet.py b/python/pyarrow/tests/test_parquet.py
index 40e44b352ac..116a0450b1b 100644
--- a/python/pyarrow/tests/test_parquet.py
+++ b/python/pyarrow/tests/test_parquet.py
@@ -265,6 +265,18 @@ def test_read_pandas_column_subset(tmpdir):
     tm.assert_frame_equal(df[['strings', 'uint8']], df_read)
 
 
+@parquet
+def test_pandas_parquet_empty_roundtrip(tmpdir):
+    df = _test_dataframe(0)
+    arrow_table = pa.Table.from_pandas(df)
+    imos = pa.BufferOutputStream()
+    _write_table(arrow_table, imos, version="2.0")
+    buf = imos.get_result()
+    reader = pa.BufferReader(buf)
+    df_read = _read_table(reader).to_pandas()
+    tm.assert_frame_equal(df, df_read)
+
+
 @parquet
 def test_pandas_parquet_pyfile_roundtrip(tmpdir):
     filename = tmpdir.join('pandas_pyfile_roundtrip.parquet').strpath