diff --git a/cpp/src/arrow/dataset/file_parquet.cc b/cpp/src/arrow/dataset/file_parquet.cc
index b5423a64b2865..1a68fe238f10e 100644
--- a/cpp/src/arrow/dataset/file_parquet.cc
+++ b/cpp/src/arrow/dataset/file_parquet.cc
@@ -103,13 +103,14 @@ static parquet::ReaderProperties MakeReaderProperties(
 }
 
 static parquet::ArrowReaderProperties MakeArrowReaderProperties(
-    const ParquetFileFormat& format, const parquet::ParquetFileReader& reader) {
+    const ParquetFileFormat& format, int64_t batch_size,
+    const parquet::ParquetFileReader& reader) {
   parquet::ArrowReaderProperties properties(/* use_threads = */ false);
   for (const std::string& name : format.reader_options.dict_columns) {
     auto column_index = reader.metadata()->schema()->ColumnIndex(name);
     properties.set_read_dictionary(column_index, true);
   }
-  properties.set_batch_size(format.reader_options.batch_size);
+  properties.set_batch_size(batch_size);
   return properties;
 }
 
@@ -351,7 +352,8 @@ Result<std::shared_ptr<Schema>> ParquetFileFormat::Inspect(
   auto properties = MakeReaderProperties(*this);
   ARROW_ASSIGN_OR_RAISE(auto reader, OpenReader(source, std::move(properties)));
 
-  auto arrow_properties = MakeArrowReaderProperties(*this, *reader);
+  auto arrow_properties =
+      MakeArrowReaderProperties(*this, parquet::kArrowDefaultBatchSize, *reader);
   std::unique_ptr<parquet::arrow::FileReader> arrow_reader;
   RETURN_NOT_OK(parquet::arrow::FileReader::Make(default_memory_pool(), std::move(reader),
                                                  std::move(arrow_properties),
@@ -368,7 +370,7 @@ Result<ScanTaskIterator> ParquetFileFormat::ScanFile(
   auto properties = MakeReaderProperties(*this, context->pool);
   ARROW_ASSIGN_OR_RAISE(auto reader, OpenReader(source, std::move(properties)));
 
-  auto arrow_properties = MakeArrowReaderProperties(*this, *reader);
+  auto arrow_properties = MakeArrowReaderProperties(*this, options->batch_size, *reader);
   return ParquetScanTaskIterator::Make(options, context, std::move(reader),
                                        std::move(arrow_properties));
 }
diff --git a/cpp/src/arrow/dataset/file_parquet.h b/cpp/src/arrow/dataset/file_parquet.h
index 752d01e7e4217..471650d4959ad 100644
--- a/cpp/src/arrow/dataset/file_parquet.h
+++ b/cpp/src/arrow/dataset/file_parquet.h
@@ -60,7 +60,6 @@ class ARROW_DS_EXPORT ParquetFileFormat : public FileFormat {
     ///
     /// @{
     std::unordered_set<std::string> dict_columns;
-    int64_t batch_size = parquet::kArrowDefaultBatchSize;
     /// @}
   } reader_options;
 
diff --git a/cpp/src/arrow/dataset/scanner.h b/cpp/src/arrow/dataset/scanner.h
index 690d1cee5c8fc..361b01ec58f4e 100644
--- a/cpp/src/arrow/dataset/scanner.h
+++ b/cpp/src/arrow/dataset/scanner.h
@@ -90,6 +90,9 @@ class ARROW_DS_EXPORT ScanOptions {
   // sub-selection optimization.
   std::vector<std::string> MaterializedFields() const;
 
+  // Maximum row count for scanned batches.
+  int64_t batch_size = 64 << 10;
+
  private:
   explicit ScanOptions(std::shared_ptr<Schema> schema);
 };
diff --git a/cpp/src/arrow/dataset/scanner_test.cc b/cpp/src/arrow/dataset/scanner_test.cc
index c683f8848c72a..804955b1311a7 100644
--- a/cpp/src/arrow/dataset/scanner_test.cc
+++ b/cpp/src/arrow/dataset/scanner_test.cc
@@ -70,6 +70,14 @@ TEST_F(TestScanner, Scan) {
   AssertScannerEqualsRepetitionsOf(MakeScanner(batch), batch);
 }
 
+TEST_F(TestScanner, DISABLED_ScanWithCappedBatchSize) {
+  // TODO(bkietz) enable when InMemory* respects ScanOptions::batch_size
+  SetSchema({field("i32", int32()), field("f64", float64())});
+  auto batch = ConstantArrayGenerator::Zeroes(kBatchSize, schema_);
+  options_->batch_size = kBatchSize / 2;
+  AssertScannerEqualsRepetitionsOf(MakeScanner(batch), batch->Slice(kBatchSize / 2));
+}
+
 TEST_F(TestScanner, FilteredScan) {
   SetSchema({field("f64", float64())});
 
diff --git a/python/pyarrow/_dataset.pyx b/python/pyarrow/_dataset.pyx
index 69bb96cf92caf..ae19bba871e18 100644
--- a/python/pyarrow/_dataset.pyx
+++ b/python/pyarrow/_dataset.pyx
@@ -110,15 +110,6 @@ cdef class ParquetFileFormatReaderOptions:
         for value in set(values):
             self.options.dict_columns.insert(tobytes(value))
 
-    @property
-    def batch_size(self):
-        """Maximum number of rows in read record batches."""
-        return self.options.batch_size
-
-    @batch_size.setter
-    def batch_size(self, int value):
-        self.options.batch_size = value
-
 
 cdef class ParquetFileFormat(FileFormat):
 
diff --git a/python/pyarrow/includes/libarrow_dataset.pxd b/python/pyarrow/includes/libarrow_dataset.pxd
index ae9e2870d38a1..ac52e00cb3384 100644
--- a/python/pyarrow/includes/libarrow_dataset.pxd
+++ b/python/pyarrow/includes/libarrow_dataset.pxd
@@ -287,7 +287,6 @@ cdef extern from "arrow/dataset/api.h" namespace "arrow::dataset" nogil:
         c_bool use_buffered_stream
         int64_t buffer_size
         unordered_set[c_string] dict_columns
-        int64_t batch_size
 
     cdef cppclass CParquetFileFormat "arrow::dataset::ParquetFileFormat"(
             CFileFormat):
diff --git a/r/src/dataset.cpp b/r/src/dataset.cpp
index 5894e42b7d435..a1e81c5cbaf54 100644
--- a/r/src/dataset.cpp
+++ b/r/src/dataset.cpp
@@ -83,9 +83,6 @@ std::shared_ptr<ds::ParquetFileFormat> dataset___ParquetFileFormat__Make(List_ o
         fmt->reader_options.dict_columns.insert(name);
       }
     }
-    if (reader_options.containsElementNamed("batch_size")) {
-      fmt->reader_options.batch_size = reader_options["batch_size"];
-    }
   }
   return fmt;
 }