diff --git a/docs/iris/src/userguide/saving_iris_cubes.rst b/docs/iris/src/userguide/saving_iris_cubes.rst
index d1658ef70e..839bfae9f2 100644
--- a/docs/iris/src/userguide/saving_iris_cubes.rst
+++ b/docs/iris/src/userguide/saving_iris_cubes.rst
@@ -4,11 +4,12 @@
 Saving Iris cubes
 ==================
 
-Iris supports the saving of cubes and cube lists to:
+Iris supports the saving of :class:~`iris.cube.Cube` and :class:~`iris.cube.CubeList` objects to:
 
 * CF netCDF (1.5)
 * GRIB (edition 2)
 * Met Office PP
+* Python pickles, via pickle module
 
 
 The :py:func:`iris.save` function saves one or more cubes to a file.
@@ -106,3 +107,38 @@ Such a custom saver will need be written to meet the needs of the file format an
 
 Implementing a bespoke saver is out of scope for the user guide.
 
+Caching cubes and cube lists to pickle files
+--------------------------------------------
+
+It should always be possible to create a temporary cache file containing a:class:~`iris.cube.Cube` or :class:~`iris.cube.CubeList` objects using the Python `Pickle <https://docs.python.org/2/library/pickle.html>`_ functionality. This can be useful when the cube or cube list has been lazily loaded so the pickle file itself will contain only a reference to the data in the original files. In this state writing, and subsequently reading, a pickle file is very fast.
+
+.. warning::
+
+    Pickle files are not a portable file format, there is no guarantee that pickle files will work if there are any differences between the python/iris environment/version.
+
+    If a pickle file contains lazily loaded cubes then any changes to the files that contain the data will break the ability of the cubes to read its data.
+
+The pickle file is especially useful for code development, where the same data files might need to be loaded repeatedly. Another common use case is parent-child processes where a single process loads data and then a number of subsequent processes can quickly access the data in order to run parallel processing tasks. Similarly the ability to pickle cubes is also vital to the using cubes in `parallel processes https://docs.python.org/2/library/multiprocessing.html`_ so any cube which cannot be pickled should be raised as an `issue on the Iris github page https://github.com/SciTools/iris/issues/new`_.
+
+A quick example of saving and reading a pickle file is:
+
+.. code-block:: python
+    import pickle
+
+    cubelist = iris.load(['a.pp', 'list.pp', 'of.pp', 'many.pp', 'files.pp'])
+    # save cubelist to a pickle
+    with open('filename.pickle', 'wb') as pfile:
+        pickle.dump(cubelist, pfile)
+
+    # then to load:
+    with open('filename.pickle', 'rb') as pfile:
+        loaded_cubelist = pickle.load(pfile)
+
+.. tip::
+    In Python 2.7, the cPickle module is a faster implementation of the pickle functionality so this will speed up the reading/writing of pickle files:
+
+    .. code-block:: python
+        # for Python 2.7:
+        import cPickle as pickle
+
+    There is no difference in Python 3.