diff -r 1fde0d70f2b8 Doc/library/gzip.rst
--- a/Doc/library/gzip.rst	Fri Nov 09 18:58:47 2012 -0800
+++ b/Doc/library/gzip.rst	Sat Nov 10 17:06:13 2012 +0100
@@ -54,6 +54,10 @@
       and the *encoding*, *errors* and *newline* arguments.
 
 
+.. class:: BadGzipfile
+
+   Is raised when encountering invalid gzip files; extends :class:`IOError`.
+ 
 .. class:: GzipFile(filename=None, mode=None, compresslevel=9, fileobj=None, mtime=None)
 
    Constructor for the :class:`GzipFile` class, which simulates most of the
diff -r 1fde0d70f2b8 Doc/whatsnew/3.4.rst
--- a/Doc/whatsnew/3.4.rst	Fri Nov 09 18:58:47 2012 -0800
+++ b/Doc/whatsnew/3.4.rst	Sat Nov 10 17:06:13 2012 +0100
@@ -149,7 +149,7 @@
 Improved Modules
 ================
 
-* None yet.
+* The :module:`gzip` module throws a specific :class:`BadGzipFile` in case of invalid or corrupted files.
 
 
 Optimizations
diff -r 1fde0d70f2b8 Lib/gzip.py
--- a/Lib/gzip.py	Fri Nov 09 18:58:47 2012 -0800
+++ b/Lib/gzip.py	Sat Nov 10 17:06:13 2012 +0100
@@ -10,7 +10,7 @@
 import builtins
 import io
 
-__all__ = ["GzipFile", "open", "compress", "decompress"]
+__all__ = ["GzipFile", "BadGzipFile", "open", "compress", "decompress"]
 
 FTEXT, FHCRC, FEXTRA, FNAME, FCOMMENT = 1, 2, 4, 8, 16
 
@@ -124,6 +124,12 @@
         return getattr(self.file, name)
 
 
+class BadGzipFile(IOError):
+    """
+    Exception raised if the processed gzip file is not valid.
+    """
+
+
 class GzipFile(io.BufferedIOBase):
     """The GzipFile class simulates most of the methods of a file object with
     the exception of the readinto() and truncate() methods.
@@ -286,10 +292,10 @@
             raise EOFError("Reached EOF")
 
         if magic != b'\037\213':
-            raise IOError('Not a gzipped file')
+            raise BadGzipFile('Not a gzipped file')
         method = ord( self.fileobj.read(1) )
         if method != 8:
-            raise IOError('Unknown compression method')
+            raise BadGzipFile('Unknown compression method')
         flag = ord( self.fileobj.read(1) )
         self.mtime = read32(self.fileobj)
         # extraflag = self.fileobj.read(1)
@@ -486,10 +492,10 @@
         crc32 = read32(self.fileobj)
         isize = read32(self.fileobj)  # may exceed 2GB
         if crc32 != self.crc:
-            raise IOError("CRC check failed %s != %s" % (hex(crc32),
-                                                         hex(self.crc)))
+            raise BadGzipFile("CRC check failed %s != %s" % 
+                              (hex(crc32), hex(self.crc)))
         elif isize != (self.size & 0xffffffff):
-            raise IOError("Incorrect length of data produced")
+            raise BadGzipFile("Incorrect length of data produced")
 
         # Gzip files can be padded with zeroes and still have archives.
         # Consume all zero bytes and set the file position to the first
diff -r 1fde0d70f2b8 Lib/test/test_gzip.py
--- a/Lib/test/test_gzip.py	Fri Nov 09 18:58:47 2012 -0800
+++ b/Lib/test/test_gzip.py	Sat Nov 10 17:06:13 2012 +0100
@@ -307,6 +307,17 @@
             d = f.read()
             self.assertEqual(d, data1 * 50, "Incorrect data in file")
 
+    def test_gzip_exception(self):
+        self.assertTrue(issubclass(gzip.BadGzipFile, IOError))
+            
+    def test_bad_gzip_file(self):
+        with open(self.filename, 'wb') as file:
+            file.write(data1 * 50)
+        
+        with gzip.GzipFile(self.filename, 'r') as file:
+            self.assertRaises(gzip.BadGzipFile, file.readlines)
+            self.assertRaises(IOError, file.readlines)
+
     def test_non_seekable_file(self):
         uncompressed = data1 * 50
         buf = UnseekableIO()