diff --git a/parquet/src/arrow/arrow_reader.rs b/parquet/src/arrow/arrow_reader.rs
index f93488f75f7..ebbb864d630 100644
--- a/parquet/src/arrow/arrow_reader.rs
+++ b/parquet/src/arrow/arrow_reader.rs
@@ -1529,8 +1529,7 @@ mod tests {
         assert_eq!(total_rows, expected_rows);
     }
 
-    #[test]
-    fn test_row_group_exact_multiple() {
+    fn test_row_group_batch(row_group_size: usize, batch_size: usize) {
         let schema = Arc::new(Schema::new(vec![Field::new(
             "list",
             ArrowDataType::List(Box::new(Field::new("item", ArrowDataType::Int32, true))),
@@ -1544,14 +1543,14 @@ mod tests {
             schema.clone(),
             Some(
                 WriterProperties::builder()
-                    .set_max_row_group_size(8)
+                    .set_max_row_group_size(row_group_size)
                     .build(),
             ),
         )
         .unwrap();
         for _ in 0..2 {
-            let mut list_builder = ListBuilder::new(Int32Builder::new(10));
-            for _ in 0..10 {
+            let mut list_builder = ListBuilder::new(Int32Builder::new(batch_size));
+            for _ in 0..(batch_size) {
                 list_builder.append(true).unwrap();
             }
             let batch = RecordBatch::try_new(
@@ -1564,9 +1563,27 @@ mod tests {
         writer.close().unwrap();
 
         let mut file_reader = ParquetFileArrowReader::try_new(Bytes::from(buf)).unwrap();
-        let mut record_reader = file_reader.get_record_reader(8).unwrap();
-        assert_eq!(8, record_reader.next().unwrap().unwrap().num_rows());
-        assert_eq!(8, record_reader.next().unwrap().unwrap().num_rows());
-        assert_eq!(4, record_reader.next().unwrap().unwrap().num_rows());
+        let mut record_reader = file_reader.get_record_reader(batch_size).unwrap();
+        assert_eq!(
+            batch_size,
+            record_reader.next().unwrap().unwrap().num_rows()
+        );
+        assert_eq!(
+            batch_size,
+            record_reader.next().unwrap().unwrap().num_rows()
+        );
+    }
+
+    #[test]
+    fn test_row_group_exact_multiple() {
+        use crate::arrow::record_reader::MIN_BATCH_SIZE;
+        test_row_group_batch(8, 8);
+        test_row_group_batch(10, 8);
+        test_row_group_batch(8, 10);
+        test_row_group_batch(MIN_BATCH_SIZE, MIN_BATCH_SIZE);
+        test_row_group_batch(MIN_BATCH_SIZE + 1, MIN_BATCH_SIZE);
+        test_row_group_batch(MIN_BATCH_SIZE, MIN_BATCH_SIZE + 1);
+        test_row_group_batch(MIN_BATCH_SIZE, MIN_BATCH_SIZE - 1);
+        test_row_group_batch(MIN_BATCH_SIZE - 1, MIN_BATCH_SIZE);
     }
 }
diff --git a/parquet/src/arrow/record_reader/mod.rs b/parquet/src/arrow/record_reader/mod.rs
index 30324fbe3e3..d2720aedeb8 100644
--- a/parquet/src/arrow/record_reader/mod.rs
+++ b/parquet/src/arrow/record_reader/mod.rs
@@ -38,7 +38,8 @@ use crate::schema::types::ColumnDescPtr;
 pub(crate) mod buffer;
 mod definition_levels;
 
-const MIN_BATCH_SIZE: usize = 1024;
+/// The minimum number of levels read when reading a repeated field
+pub(crate) const MIN_BATCH_SIZE: usize = 1024;
 
 /// A `RecordReader` is a stateful column reader that delimits semantic records.
 pub type RecordReader<T> =