diff --git a/Cargo.toml b/Cargo.toml
index 87ac73f1a49..eb1c1017ba6 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -69,7 +69,7 @@ rand = { version = "0.8", optional = true }
 
 itertools = { version = "^0.10", optional = true }
 
-base64 = { version = "0.13.0", optional = true }
+base64 = { version = "0.21.0", optional = true }
 
 # to write to parquet as a stream
 futures = { version = "0.3", optional = true }
diff --git a/src/io/parquet/read/schema/metadata.rs b/src/io/parquet/read/schema/metadata.rs
index 2595083f44e..1df6e7b2af2 100644
--- a/src/io/parquet/read/schema/metadata.rs
+++ b/src/io/parquet/read/schema/metadata.rs
@@ -1,3 +1,4 @@
+use base64::{engine::general_purpose, Engine as _};
 pub use parquet2::metadata::KeyValue;
 
 use crate::datatypes::{Metadata, Schema};
@@ -18,7 +19,7 @@ pub fn read_schema_from_metadata(metadata: &mut Metadata) -> Result<Option<Schem
 
 /// Try to convert Arrow schema metadata into a schema
 fn get_arrow_schema_from_metadata(encoded_meta: &str) -> Result<Schema> {
-    let decoded = base64::decode(encoded_meta);
+    let decoded = general_purpose::STANDARD.decode(encoded_meta);
     match decoded {
         Ok(bytes) => {
             let slice = if bytes[0..4] == [255u8; 4] {
diff --git a/src/io/parquet/write/schema.rs b/src/io/parquet/write/schema.rs
index 248397d2414..ee60e98ac13 100644
--- a/src/io/parquet/write/schema.rs
+++ b/src/io/parquet/write/schema.rs
@@ -1,3 +1,4 @@
+use base64::{engine::general_purpose, Engine as _};
 use parquet2::{
     metadata::KeyValue,
     schema::{
@@ -30,7 +31,7 @@ pub fn schema_to_metadata_key(schema: &Schema) -> KeyValue {
     len_prefix_schema.extend_from_slice(&(schema_len as u32).to_le_bytes());
     len_prefix_schema.extend_from_slice(&serialized_schema);
 
-    let encoded = base64::encode(&len_prefix_schema);
+    let encoded = general_purpose::STANDARD.encode(&len_prefix_schema);
 
     KeyValue {
         key: ARROW_SCHEMA_META_KEY.to_string(),