It is rare for a story to have only a single narrative level: in fact, even the simplest stories usually contain multiple nested stories. The following is an annotation guide for encoding the boundaries between narrative levels, and which has been validated on modern fiction and TV scripts. We provided definitions of and give instructions for annotating information about each narrative level: embedded narratives, interruptive narrative, flashbacks (analepsis), and flashforwards (prolepsis). This annotation schema can be used for many types of narratological and computational research, however our intention in developing it was to lay the foundation for training computers to automatically extract narrative levels from long text.