Unlike a PostMortem, you can do this one at any time in the project. Just time-travel in imagination to any date after the scheduled release date, and accept as a premise that the project has failed miserably. Now come up with an answer to the question "what happened". If any of your answers make sense, you have not only pinpointed a project risk, you also have a vivid description of how it could be fatal to the project. Acting on that (i.e. adopting a risk resolution strategy) has just become a lot easier. -- LaurentBossavit
----
It might happen that your analysis is hindered by the effort of thinking the unthinkable. I.e the effort to keep the premise accepted during all the time of  the reflexion. Just thinking about some current project, many possible causes of death come to my mind. I guess I should focus on the most lethal cause. Then again, is there a life before death ? My project is not really a living one as a matter of fact (I mean the diastole/systole cycles are so long). --ChristopheThibaut.
----
When developing software, I often have episodes where I've written code to implement some feature, I run a test, and it fails.  Before I even look at the error message, it occurs to me "Oh, yeah!  I forgot to ..."  I've often wondered if the same phenomenon would work if I just pretended that it failed, before running the test, and whether it would work at a project level rather than just at a micro-iteration level.

When performing this exercise at a project level, it is probably useful to assume one or more particular types of miserable failure: over budget, behind schedule, too buggy, incompatible with customers' systems, inferior to the competition, the entire team quit, etc.  Thinking of the types of potential failures leads one to think about which types of failure are most likely and which ones are worse than the others.  This helps to prioritize risks.  --KrisJohnson
----
ugh, terrible name... it directly implies "You are going to die."

Alternate names being used in the industry are IterationRetrospective, PrePerspective (a Retrospective held '''before''' the project), AdaptiveRetrospective, MilestoneRetrospective and ReflectionWorkshop, all better names than PreMortem.
----
If it appears highly likely that death of the project will occur, one might consider employing an AbandonmentStrategy.
----
See also TopTenRisks