You need to do a Risk Assessment. What functionality is most critical to your end-user. Those test cases are more important. Within those pick the ones that will likely be most used and produce the most realistic usage of the system by the end-user. These get bumped up higher on the list. Of those which ones will give you the most coverage of the critical functionality. This is in affect using the 80/20 rule. Exercise the 20% of an application that is used 80% of the time.
It isn't perfect, but in certain situations it gives you a fighting chance.
I assign risk assessments in the Test Plan, generally not down to the individual test case level but more at the functional area (read: Test Procedure) level. The Test Plan is reviewed by all the appropriate people as defined by our process, and they have the opportunity to review my risk assessments and approve them or recommend changes. (Historically, they seldom change them - either because my risk assessments are good, or perhaps because they don't really read them???)
PS: Just because a Test Plan gets signed off does not mean those risk levels are set in stone. There always needs to be open lines of communication with all the development teams and other stakeholders to determine if priorities/risks need to be changed.
"Any sufficiently advanced bug is indistinguishable from a feature." -Bruce Brown
[This message has been edited by Charles Reace (edited 11-21-2002).]
If your lucky the group will decide, but you should do a first run yourself. Give them something to look at and then iron out any differences. What you may think is important may not be to the other people. Group consensus is a good thing to have.
Also, if you do get consensus and buy in then later if the stuff hits the fan you have documentation stating that it was a group decision and that people were informed.
I did this a couple of times with tight schedules and when certain people wanted to change direction or throw something else in at the last minute it was an effective way of keeping things under some semblance of control.