From http://www.fmsfonline.org/hypnosis.html#chbf

Experimental data indicate that hypnosis has three main effects upon memory.  Hypnosis increases productivity, but most of the new information is in error;  it increases confidence for both correct and incorrect "novel" remembrances;  these increases in productivity and confidence are found at all levels of hypnotizability, but these effects are most pronounced in high hypnotizables in hypnosis (as compared to when they are assigned to an imagination or a repeated recall condition), and as compared to low hypnotizables who receive either hypnosis, imagination and repeated recall instructions (Nogrady, McConkey & Perry, 1985; Orne, Soskis, Dinges, Orne & Tonry, 1985).

Hypnosis can result in confabulation. This is the tendency to confuse fantasy as fact. Again, as was emphasized earlier, the possibility that novel information elicited in hypnosis may be confabulated must be evaluated -- always. It is possible that such new information is true, but as equally, it could be a lie, it could be confabulated, or it could be pseudo-memory that is manufactured in response to the demands of the hypnotic situation. These four alternatives -- truth, lie, confabulation or pseudo-memory -- were emphasized by French forensic investigators of over a century ago (Laurence & Perry, 1988), and each of them, still, requires rigorous assessment.

"Hypermnesia" refers to an abnormally vivid or complete memory. Hypnotic hypermnesia is the belief that hypnosis enhances accurate memory for events that a person is initially unable to remember. In actuality, there is little evidence favoring the hypnotic hypermnesia effect. Although people may produce more information with the use of hypnosis, it is not necessarily accurate. Given that the processes underlying response to hypnosis appear to implicate fantasy and imagination, any novel material elicited in hypnosis needs to be corroborated by independent means. This is particularly so when hypnosis is employed as a part of a police investigation designed to elicit additional leads. Uncorroborated, hypnotically elicited memories can, all too easily, lead to the wrongful imprisonment of innocent people.

Post hypnotic amnesia is the failure to remember most, if not all of the events occurring in hypnosis, until a pre-arranged signal to recall them is administered. This phenomenon is confined to the top 10-15% of the population, and, typically, these individuals remember mere fragments of what transpired during the preceding period of hypnosis. They may, for instance, recall writing their name, their age, and the date, but not that this request was made within the context of being hypnotically age regressed to childhood. While some high hypnotizables have a "blanket" amnesia for the events of hypnosis, this is more typical of low hypnotizables asked to simulate hypnosis (Orne, 1979). Usually, the response of simulators to this and other hypnotic items is to respond in terms of the perceived demands of the situation. They tend to interpret an amnesia suggestion as meaning that they cannot remember anything at all until they receive the signal to reverse the amnesia.

This reversibility of post-hypnotic amnesia distinguishes it from amnesias of organic origin, such as from a blow to the head. While evidence suggests that memories based upon a retrograde amnesia (that is, one resulting from trauma of either a psychological or physical character) may become available eventually, the retrieval process is, ordinarily, slow and laborious. By contrast, the reversal of hypnotically suggested amnesias is effected by a simple suggestion that the person "can now remember everything."

right on man