diff --git a/scikits/learn/utils/crossval.py b/scikits/learn/utils/crossval.py
index b4704c4470f57db5861e9b4183ee14d824213cd8..0669e0e2c2a2233bf25a2e5c0fdbc9a949523370 100644
--- a/scikits/learn/utils/crossval.py
+++ b/scikits/learn/utils/crossval.py
@@ -3,8 +3,6 @@
 
 # $Id$
 
-import exceptions
-
 import numpy as np
 
 try:
@@ -136,14 +134,14 @@ def k_fold(n, k):
         yield train_index, test_index
 
 
-def leave_one_label_out(label):
+def leave_one_label_out(labels):
     """
     Leave-One-Label_Out cross validation:
     Provides train/test indexes to split data in train test sets
 
     Parameters
     ----------
-    label : list
+    labels : list
             List of labels
 
     Examples
@@ -152,8 +150,8 @@ def leave_one_label_out(label):
     >>> from scikits.learn.utils import crossval
     >>> X = [[1, 2], [3, 4], [5, 6], [7, 8]]
     >>> y = [1, 2, 1, 2]
-    >>> label = [1,1,2,2]
-    >>> lol = crossval.leave_one_label_out(label)
+    >>> labels = [1, 1, 2, 2]
+    >>> lol = crossval.leave_one_label_out(labels)
     >>> for train_index, test_index in lol:
     ...    print "TRAIN:", train_index, "TEST:", test_index
     ...    X_train, X_test, y_train, y_test = crossval.split(train_index, \
@@ -169,9 +167,9 @@ def leave_one_label_out(label):
      [7 8]] [1 2] [1 2]
 
     """
-    for i in np.unique(label):
-        test_index  = np.zeros(len(label), dtype=np.bool)
-        test_index[label==i] = True
+    for i in np.unique(labels):
+        test_index  = np.zeros(len(labels), dtype=np.bool)
+        test_index[labels==i] = True
         train_index = np.logical_not(test_index)
         yield train_index, test_index