This overview briefly describes what the config classes are and what they are for. All the details about how to use them may be found in the description of the wxConfigBase class and the documentation of the file, registry and INI file based implementations mentions all the features/limitations specific to each one of these versions.
The config classes provide a way to store some application configuration information. They were especially designed for this usage and, although may probably be used for many other things as well, should be limited to it. It means that this information should be:
On the other hand, the features provided make them very useful for storing all kinds of small to medium volumes of hierarchically-organized, heterogeneous data. In short, this is a place where you can conveniently stuff all your data (numbers and strings) organizing it in a tree where you use the filesystem-like paths to specify the location of a piece of data. In particular, these classes were designed to be as easy to use as possible.
From another point of view, they provide an interface which hides the differences between the Windows registry and the standard Unix text format configuration files. Other (future) implementations of wxConfigBase might also understand GTK resource files or their analogues on the KDE side.
In any case, each implementation of wxConfigBase does its best to make the data look the same way everywhere. Due to limitations of the underlying physical storage, it may not implement 100% of the base class functionality.
There are groups of entries and the entries themselves. Each entry contains either a string or a number (or a boolean value; support for other types of data such as dates or timestamps is planned) and is identified by the full path to it: something like /MyApp/UserPreferences/Colors/Foreground. The previous elements in the path are the group names, and each name may contain an arbitrary number of entries and subgroups. The path components are always separated with a slash, even though some implementations use the backslash internally. Further details (including how to read/write these entries) may be found in the documentation for wxConfigBase.