Yesterday, a New Jersey jury added $9 million in punitive damages to $4.5 million in compensation it had awarded to John McDarby, a 77-year-old man who had sued the drug manufacturer Merck after suffering a heart attack while taking the pain reliever Vioxx. That verdict did not tell us much about Vioxx, Merck, and even the Food and Drug Administration, but it certainly provided a lot of unpleasant news about the tort system.

Essentially, the jury had to decide three things. One was whether Vioxx was a substantially contributing factor to the heart attacked suffered by an elderly ex-smoker with diabetes and coronary artery disease. Unfortunately, science can’t give us a good answer. Mr. McDarby was clearly at great risk for a heart attack. But there is no way to know which of a multitude of precipitating factors tipped the balance on that terrible day. We don’t even have reliable data indicating an extra probability of a heart attack in the relevant circumstances. The plaintiff lawyers simply dismissed the fact that after combing through a massive literature, the FDA concluded that traditional anti-inflammatories like Advil and old prescription drugs (called nonsteroidal anti-inflammatory drugs), as well as Vioxx and its modern competitors (like Celebrex), all probably involve a heart attack risk which remained largely undiscovered until large-scale clinical trials were run on Vioxx and its competitors.

The jury simply had no way to determine whether the plaintiff would have had a heart attack if he had used a different pain reliever. Relying on an unpredictable mix of intuition and emotion, the jury simply guessed.

The second issue was failure-to-warn. The plaintiff argued that in addition to complying with all FDA regulations about warnings, Merck should have specifically warned the doctor not to prescribe Vioxx to this patient. What warning, we should ask, and to what effect? Here, the jury had even less foundation than when deciding causation. The most important data about Vioxx and heart attacks had long since been published and widely discussed in medical circles. Moreover, Vioxx was a lifesaver for patients with a propensity toward fatal ulcers, which every year kill thousands of patients taking older drugs that Vioxx replaced. Whether more warnings would have pushed the patient toward more safety or just rearranged Mr. McDarby’s numerous various risk factors was something the jury could not possibly discern. No scientifically based warning could have said something like “there is a so-and-so percent [“X% better?] extra chance of a heart attack.” A warning would presumably have said something like “don’t prescribe this to patients with a high risk of heart attack,” but what good would that have done in the absence of information about whether the obvious alternative treatments were better?

Again, the jury had to guess. And again, there is no reason to think its guess would have been better than the FDA’s own expert assessment.

The third question was whether to award punitive damages, something never before done to a pharmaceutical manufacturer liability case in New Jersey. Under New Jersey law, the jury had to find that Merck had withheld information from the FDA in a “wanton and willful” manner. Again, a problem for the jury was the FDA itself. As a former FDA staffer testified, the FDA was satisfied that Merck had turned over all the information it needed. How did the jury reach a conclusion so different from the FDA’s own? Here lies one of the most unsettling parts of this trial. Merck, like all responsible manufacturers, massaged the Vioxx data as it arrived, trying to get a fix on the true heart-attack risk and other matters in a diverse mix of patients. The FDA has to know that firms do that kind of analysis all the time. The jury said Merck should have shared its own preliminary assessment of the data with the FDA even though the FDA famously makes up its own mind about such things and the underlying data had in fact already been given to the FDA. Had Merck not undertaken its analysis, there would have been nothing to keep from the FDA. The lesson for Merck and the rest of the industry is that mulling over safety data can get you in legal trouble even when the FDA has no interest in the hundreds of spreadsheet calculations that necessarily arise.

Now, if we ask a jury to do the impossible, we’re asking for trouble. And if we ask it to do the impossible with tens of billions of dollars in the balance, we’re asking for big trouble. Yet that’s what we did last week in that New Jersey courtroom. Does this mean we should junk the liability system, at least for pharmaceuticals? Not at all. The system works reasonably well for outlier situations such as intentionally hiding essential information from the FDA and others, or marketing drugs that clearly fail a cost-benefit test while withholding exactly the information that would have disclosed that fact. But today’s system cannot work well for close calls, at least not when the stakes are in the billions or tens or hundreds of billions of dollars. Analysts expect tens of thousands of cases roughly similar to Mr. McDarby’s. Juries will continue to guess at the central issues, getting it wrong as often as right but charging Merck $10 million or $20 million or $40 million every time they’re wrong. A thousand mistakes is $10 billion to $40 billion. And a lot of states make it easier than New Jersey does to award punitive damages.

Nor is Vioxx the only drug where this perverse logic could play out. The liability plaintiff bar is honing the same methods for other drugs where similar imponderables can generate $10 million guesses.

What to do? There is a simple way to keep these questions out of the reach of juries. The states can do what the FDA has already suggested they do: Stop letting plaintiff lawyers seek huge awards for behavior that actually complies with the FDA’s regulations and findings of fact. With lawyers bringing thousands cases before juries that read about the evils of the drug companies every day and then have to have to decide the central issues through guesswork, we are now creating a monster. It is time to rein that monster in.

Mr. Calfee is a resident scholar at the American Enterprise Institute, which receives a small portion of its funding from pharmaceutical firms.