diff --git a/python/pyarrow/_flight.pyx b/python/pyarrow/_flight.pyx
index 8289215de2e29..78c518661e14b 100644
--- a/python/pyarrow/_flight.pyx
+++ b/python/pyarrow/_flight.pyx
@@ -1743,8 +1743,9 @@ cdef class RecordBatchStream(FlightDataStream):
     cdef:
         object data_source
         CIpcWriteOptions write_options
+        int64_t max_chunksize
 
-    def __init__(self, data_source, options=None):
+    def __init__(self, data_source, options=None, max_chunksize=None):
         """Create a RecordBatchStream from a data source.
 
         Parameters
@@ -1753,6 +1754,9 @@ cdef class RecordBatchStream(FlightDataStream):
             The data to stream to the client.
         options : pyarrow.ipc.IpcWriteOptions, optional
             Optional IPC options to control how to write the data.
+        max_chunksize : int, default None
+            Optional maximum number of rows for each chunk.
+            Only applicable if the data source is a Table.
         """
         if (not isinstance(data_source, RecordBatchReader) and
                 not isinstance(data_source, lib.Table)):
@@ -1760,6 +1764,10 @@ cdef class RecordBatchStream(FlightDataStream):
                             "but got: {}".format(type(data_source)))
         self.data_source = data_source
         self.write_options = _get_options(options).c_options
+        if max_chunksize is not None:
+            self.max_chunksize = max_chunksize
+        else:
+            self.max_chunksize = 0
 
     cdef CFlightDataStream* to_stream(self) except *:
         cdef:
@@ -1768,7 +1776,10 @@ cdef class RecordBatchStream(FlightDataStream):
             reader = (<RecordBatchReader> self.data_source).reader
         elif isinstance(self.data_source, lib.Table):
             table = (<Table> self.data_source).table
-            reader.reset(new TableBatchReader(deref(table)))
+            batch_reader = new TableBatchReader(deref(table))
+            if self.max_chunksize > 0:
+                batch_reader.set_chunksize(self.max_chunksize)
+            reader.reset(batch_reader)
         else:
             raise RuntimeError("Can't construct RecordBatchStream "
                                "from type {}".format(type(self.data_source)))