diff --git a/torchvision/datasets/folder.py b/torchvision/datasets/folder.py
index 633ebab0153..041874bacbc 100644
--- a/torchvision/datasets/folder.py
+++ b/torchvision/datasets/folder.py
@@ -23,7 +23,7 @@ def find_classes(dir):
 
 def make_dataset(dir, class_to_idx):
     images = []
-    for target in os.listdir(dir):
+    for target in class_to_idx.keys():
         d = os.path.join(dir, target)
         if not os.path.isdir(d):
             continue
@@ -43,10 +43,33 @@ def default_loader(path):
 
 
 class ImageFolder(data.Dataset):
-
-    def __init__(self, root, transform=None, target_transform=None,
-                 loader=default_loader):
-        classes, class_to_idx = find_classes(root)
+    """
+    A class representing a directory of images as a `Dataset`.
+
+    Args:
+        root (String): The path to the directory.
+        transform (Object): A callable object that transforms images. See: torchvision/transforms.py. By default no
+                            transformations will be applied to images.
+        target_transform (Object): A callable object that transforms targets (labels). By default no transformations
+                                   will be applied to targets.
+        loader (Function): Loads an image and returns it in a usable form. By default loads images by
+                           their path and returns a `PIL.Image` instance.
+        classes (List/Tuple): The sub-directories of `root` that correspond to the classes of this data set. By default
+                              all sub-directories of `root` are used.
+
+    Example:
+        >>> dataset = folder.ImageFolder('./dataset', transform=transforms.Compose([
+        >>>     transforms.Scale(size=224),
+        >>>     transforms.RandomCrop(size=224),
+        >>>     transforms.ToTensor()
+        >>> ]), classes=['cat', 'dog'])
+    """
+
+    def __init__(self, root, transform=None, target_transform=None, loader=default_loader, classes=None):
+        if not classes:
+            classes, class_to_idx = find_classes(root)
+        else:
+            class_to_idx = {classes[i]: i for i in range(len(classes))}
         imgs = make_dataset(root, class_to_idx)
         if len(imgs) == 0:
             raise(RuntimeError("Found 0 images in subfolders of: " + root + "\n"