I've had several people ask me why they keep seeing different numbers for survival percentages. Here's the heart of the problem:

It all depends upon what you mean by survival. In days gone by - and for very good reasons in the modern age - 'survival' meant the "return of spontaneous circulation" (Come on, folks, couldn't we just say the victim got his or her pulse back?)
Unfortunately, getting your pulse back doesn't always leave you at the same station.

The difference is your CPC score:

CPC = 1 means you have no significant neurological deficit.
CPC = 2 means you are able to perform the activities of daily living, perhaps with some minimal accommodation.
CPC = 3 means you have some severe neurological deficit that interferes to varying degrees with your ability to perform the activities of daily living.
CPC = 4 means you are in a coma.

The percentage of the victims that survive depends upon what you mean by 'survive':

The CDC's publication of five years of CARES data underscores the importance of time.

On a sample of 31,583 cardiac arrest victims, a Bystander applied the AED only 3.7% of the time, and the 911 Responder applied the AED the other 96.7% of the time. Twenty-three percent of the people to whom a Bystander applied an AED survived, and nine percent of the people to whom the 911 Responder applied the AED survived.

This doesn't mean that Bystanders are two and a half times better at applying AEDs. It means that the Bystander was able to get an AED on the victim sooner, if there were one there. (These numbers won't tie exactly to the CARES ROSC data because they have been adjusted to eliminate the victims who didn't die but who weren't independently able to perform the activities of daily living after the incident.)