There are several different ways to classify long-term memories according to their content.
Declarative memory is a term for information which is available to conscious recollection and verbal retrieval (i.e., it can be "declared"). Two subclasses of declarative memory are episodic memory, which is autobiographical information, and semantic memory, which is factual information about the world (vocabulary items, knowledge of what a hammer is used for, memory of multiplication tables, etc.).

Non-declarative memory is a catch-all phrase for all the other kinds of memory which are less accessible to conscious recollection and verbal retrieval. Examples are skills and habits. For example, you may know how to tie your shoelaces or how to ride a bicycle -- but if asked to verbalize how you do these things, it is difficult. Non-declarative memory is sometimes called procedural memory or "knowing how" to do something, as distinct from declarative memory which is "knowing that" something is true. Patients with anterograde amnesia, a memory disorder affecting the ability to form new memories, often have little ability to form new declarative memories but may be relatively spared in the ability to form new non-declarative memories.

Another way to classify long-term memory is as explicit memory versus implicit memory. Explicit memories are memories that you "know" you know. For example, if someone asks you for your name, you "know" that you know the answer. Implicit memories are memories that you may not have been aware of acquiring; it may be only when you are tested for this information that you realize it is there. Implicit memory is of special interest when studying patients with brain damage that affects memory; sometimes, these patients may learn things without being aware -- i.e., explicit memory may be damaged but implicit memory may be spared. In this case, the way that memory is assessed can have a great effect on how accurately the patient is diagnosed.

Memory can also be assessed by testing recall or recognition. In a recall test, the examiner might read a list of words aloud, and then as the patient to recall as many words as possible. Alternately, the examiner might show the patient a second list of words and ask the patient to recognize which of those words were on the original list. Typically, people are much more accurate on tests of recognition than on tests of recall. This is probably because the recognition test provides cues (the list of words to be recognized) which can help prompt memory. Patients with anterograde amnesia, a deficit in new memory formation, may only be able to generate a few words on a recall test, but may do almost as well as healthy people on a recognition test.