diff --git a/amlb/datasets/file.py b/amlb/datasets/file.py
index bcb941ab2..8e696f135 100644
--- a/amlb/datasets/file.py
+++ b/amlb/datasets/file.py
@@ -15,7 +15,7 @@
 from ..resources import config as rconfig
 from ..utils import Namespace as ns, as_list, lazy_property, list_all_files, memoize, path_from_split, profile, split_path
 
-from .fileutils import download_file, is_archive, is_valid_url, unarchive_file, url_exists
+from .fileutils import is_archive, is_valid_url, unarchive_file, get_file_handler
 
 
 log = logging.getLogger(__name__)
@@ -118,8 +118,9 @@ def _extract_train_test_paths(self, dataset, fold=None):
         elif is_valid_url(dataset):
             cached_file = os.path.join(self._cache_dir, os.path.basename(dataset))
             if not os.path.exists(cached_file):  # don't download if previously done
-                assert url_exists(dataset), f"Invalid path/url: {dataset}"
-                download_file(dataset, cached_file)
+                handler = get_file_handler(dataset)
+                assert handler.exists(dataset), f"Invalid path/url: {dataset}"
+                handler.download(dataset, dest_path=cached_file)
             return self._extract_train_test_paths(cached_file)
         else:
             raise ValueError(f"Invalid dataset description: {dataset}")
diff --git a/amlb/datasets/fileutils.py b/amlb/datasets/fileutils.py
index bce8cbc4f..9be9a84e3 100644
--- a/amlb/datasets/fileutils.py
+++ b/amlb/datasets/fileutils.py
@@ -2,39 +2,85 @@
 import os
 import shutil
 import tarfile
+import boto3
+from botocore.errorfactory import ClientError
 from urllib.error import URLError
 from urllib.parse import urlparse
-from urllib.request import Request, urlopen, urlretrieve
+from urllib.request import Request, urlopen
 import zipfile
 
 from ..utils import touch
 
 log = logging.getLogger(__name__)
 
-SUPPORTED_SCHEMES = ("http", "https")
 
+class FileHandler:
+    def exists(self, url): pass
+    def download(self, url, dest_path): pass
+
+
+class HttpHandler(FileHandler):
+    def exists(self, url):
+        head_req = Request(url, method='HEAD')
+        try:
+            with urlopen(head_req) as test:
+                return test.status == 200
+        except URLError as e:
+            log.error(f"Cannot access url %s: %s", url, e)
+            return False
+    
+    def download(self, url, dest_path):
+        touch(dest_path)
+        with urlopen(url) as resp, open(dest_path, 'wb') as dest:
+            shutil.copyfileobj(resp, dest)
+
+
+class S3Handler(FileHandler):
+    def exists(self, url):
+        s3 = boto3.client('s3')
+        bucket, key = self._s3_path_to_bucket_prefix(url)
+        try:
+            s3.head_object(Bucket=bucket, Key=key)
+            return True
+        except ClientError as e:
+            log.error(f"Cannot access url %s: %s", url, e)
+            return False
+        
+    def download(self, url, dest_path):
+        touch(dest_path)
+        s3 = boto3.resource('s3')
+        bucket, key = self._s3_path_to_bucket_prefix(url)
+        try:
+            s3.Bucket(bucket).download_file(key, dest_path)
+        except ClientError as e:
+            if e.response['Error']['Code'] == "404":
+                log.error("The object does not exist.")
+            else:
+                raise
+        
+    def _s3_path_to_bucket_prefix(self, s3_path):
+        s3_path_cleaned = s3_path.split('://', 1)[1]
+        bucket, prefix = s3_path_cleaned.split('/', 1)
+        return bucket, prefix
 
-def is_valid_url(url):
-    return urlparse(url).scheme in SUPPORTED_SCHEMES
 
+scheme_handlers = dict(
+    http=HttpHandler(),
+    https=HttpHandler(),
+    s3=S3Handler(),
+    s3a=S3Handler(),
+    s3n=S3Handler(),
+)
 
-def url_exists(url):
-    if not is_valid_url(url):
-        return False
-    head_req = Request(url, method='HEAD')
-    try:
-        with urlopen(head_req) as test:
-            return test.status == 200
-    except URLError as e:
-        log.error(f"Cannot access url %s: %s", url, e)
-        return False
+SUPPORTED_SCHEMES = list(scheme_handlers.keys())
 
 
-def download_file(url, dest_path):
-    touch(dest_path)
-    # urlretrieve(url, filename=dest_path)
-    with urlopen(url) as resp, open(dest_path, 'wb') as dest:
-        shutil.copyfileobj(resp, dest)
+def get_file_handler(url):
+    return scheme_handlers[urlparse(url).scheme]
+
+
+def is_valid_url(url):
+    return urlparse(url).scheme in SUPPORTED_SCHEMES
 
 
 def is_archive(path):
@@ -52,4 +98,3 @@ def unarchive_file(path, dest_folder=None):
         with tarfile.open(path) as tf:
             tf.extractall(path=dest_folder)
     return dest
-