diff --git a/examples/parquet_read_async.rs b/examples/parquet_read_async.rs
index 2428b368869..3f1abc4faf6 100644
--- a/examples/parquet_read_async.rs
+++ b/examples/parquet_read_async.rs
@@ -43,7 +43,7 @@ async fn main() -> Result<()> {
         // the runtime.
         // Furthermore, this operation is trivially paralellizable e.g. via rayon, as each iterator
         // can be advanced in parallel (parallel decompression and deserialization).
-        let chunks = RowGroupDeserializer::new(column_chunks, row_group.num_rows() as usize, None);
+        let chunks = RowGroupDeserializer::new(column_chunks, row_group.num_rows(), None);
         for maybe_chunk in chunks {
             let chunk = maybe_chunk?;
             println!("{}", chunk.len());
diff --git a/examples/s3/src/main.rs b/examples/s3/src/main.rs
index 5aa24431dd9..d3f668d4421 100644
--- a/examples/s3/src/main.rs
+++ b/examples/s3/src/main.rs
@@ -71,7 +71,7 @@ async fn main() -> Result<()> {
 
     // this is CPU-bounded and should be sent to a separate thread-pool.
     // We do it here for simplicity
-    let chunks = read::RowGroupDeserializer::new(column_chunks, group.num_rows() as usize, None);
+    let chunks = read::RowGroupDeserializer::new(column_chunks, group.num_rows(), None);
     let chunks = chunks.collect::<Result<Vec<_>>>()?;
 
     // this is a single chunk because chunk_size is `None`
diff --git a/src/io/parquet/read/row_group.rs b/src/io/parquet/read/row_group.rs
index 4dde21e467d..59fba886a65 100644
--- a/src/io/parquet/read/row_group.rs
+++ b/src/io/parquet/read/row_group.rs
@@ -260,8 +260,6 @@ pub async fn read_columns_many_async<
     field_columns
         .into_iter()
         .zip(fields.into_iter())
-        .map(|(columns, field)| {
-            to_deserializer(columns, field, row_group.num_rows() as usize, chunk_size)
-        })
+        .map(|(columns, field)| to_deserializer(columns, field, row_group.num_rows(), chunk_size))
         .collect()
 }
diff --git a/tests/it/io/parquet/read_indexes.rs b/tests/it/io/parquet/read_indexes.rs
index 68f09bae804..328a826eaf0 100644
--- a/tests/it/io/parquet/read_indexes.rs
+++ b/tests/it/io/parquet/read_indexes.rs
@@ -125,7 +125,7 @@ fn read_with_indexes(
         vec![&c1.descriptor().descriptor.primitive_type],
         schema.fields[1].clone(),
         None,
-        row_group.num_rows() as usize,
+        row_group.num_rows(),
     )?;
 
     let arrays = arrays.collect::<Result<Vec<_>>>()?;
diff --git a/tests/it/io/parquet/write_async.rs b/tests/it/io/parquet/write_async.rs
index 5644caf67f5..000c39ca64c 100644
--- a/tests/it/io/parquet/write_async.rs
+++ b/tests/it/io/parquet/write_async.rs
@@ -63,7 +63,7 @@ async fn test_parquet_async_roundtrip() {
         let column_chunks = read_columns_many_async(factory, group, schema.fields.clone(), None)
             .await
             .unwrap();
-        let chunks = RowGroupDeserializer::new(column_chunks, group.num_rows() as usize, None);
+        let chunks = RowGroupDeserializer::new(column_chunks, group.num_rows(), None);
         let mut chunks = chunks.collect::<Result<Vec<_>>>().unwrap();
         out.append(&mut chunks);
     }