Old-fashioned CPR is getting a makeover. Cardiopulmonary resuscitation (search) is crucial when people collapse with cardiac arrest, but it's hard to perform correctly.

Now major efforts are under way to improve how doctors, paramedics and average bystanders do the job: New CPR guidelines are due this fall, and high-tech machines that promise to help are already showing up in ambulances and offices.

Not yet proven is whether using technology — like a chest-squeezing gadget or sensors that coax rescuers to pound harder — to spice up the 40-year-old resuscitation technique really will save lives.

Emergency-care specialists agree that CPR today doesn't save as many lives as it could.

"We've got our work cut out for us to make sure CPR is done better," says Mary Fran Hazinski of the American Heart Association, which is finalizing new recommendations designed to do just that.

More than 300,000 Americans each year die of cardiac arrest, where the heart's electrical system goes haywire and the heart abruptly stops beating.

Portable defibrillators can increase survival, delivering a jolt of electricity that stuns the heart, ending the abnormal rhythm and giving it a chance to resume a normal beat.

But the heart-zappers alone aren't enough. Virtually all cardiac-arrest victims need CPR, too. It buys time until a defibrillator arrives. Often, it's needed immediately after zapping, as the heart struggles to resume circulation.

Also, studies show that doing CPR first makes defibrillation more likely to work if cardiac arrest has lasted longer than three minutes. The longer someone goes without oxygen, the more their abnormal heart rhythm degrades until it's unshockable.

But "it has to be good CPR. We don't want to delay defibrillation for crummy CPR," warns Dr. Lance Becker of the University of Chicago, co-author of one of a pair of surprising studies earlier this year that found even the best-trained rescuers — doctors, nurses and paramedics — too frequently give inadequate CPR.

The studies found long pauses in CPR; that rescuers often didn't pound hard or fast enough on victims' chests; and that they pumped too much air into the lungs (a mistake more prone to professionals using hand-held air bags instead of mouth-to-mouth breathing.)

Why? Good CPR is tough — you must compress the chest 1 1/2 to 2 inches deep — and rescuers tire or may pause to prepare the defibrillator or perform other tasks.

Enter the high-tech machines:

--Philips Medical Systems is seeking Food and Drug Administration clearance of a CPR-aiding defibrillator with sensor pads that measure chest compressions. The Q-CPR system's recorded voice says "compress deeper," or faster, or slower, or not so deep. Additional sensors also measure whether lungs are being filled too fast or slow. Initially aimed at paramedics, Philips plans to put the system on at-home defibrillators soon, too.

--Zoll Medical Corp. already sells the AED-Plus defibrillator with a chest-measuring sensor that beeps to let rescuers pace compressions.

--And more than 100 fire departments and other first responders have bought Zoll's AutoPulse, a battery-powered band that squeezes the chest to replace manual compressions.

It's getting some rave reviews.

"CPR is exhausting," explains Glenn Ortiz-Schuldt, the San Francisco Fire Department's chief of emergency medical services. "The machine doesn't tire."

A recent study of his fire department found more AutoPulse recipients made it to the hospital alive than those getting manual CPR — 39 percent compared with 29 percent.

The key question is whether those people also were more likely to recover without brain damage and leave the hospital. The University of Washington has begun the first study, called ASPIRE, to find out.

But researchers recently suspended work because manual CPR seemed to work better at one of the five study sites. Another snag: Officials in Riverside, Calif., suspended AutoPulse use because the coroner was suspicious of injuries sustained by a 77-year-old man who died despite its use; Zoll says the machine isn't to blame.

The technology's intriguing, says the heart association's Hazinski, but no machine yet — and there have been previous attempts — has proven better than manual CPR. New international guidelines due in November aim to teach rescuers to do CPR better.

But the average person doesn't have to wait. Take a course, says Hazinski — and take another one if you learned years ago.

"You have to learn it and practice it, and not be afraid to push hard."