SHOULD pharmaceutical companies have to reveal the results of clinical trials they conduct on their drugs, even when the results show the drugs to be ineffective? That's the issue behind a discussion that has begun among the editors of the nation's medical journals. Concerned that drug companies may be sending them only partial results from their clinical trials, they now want to set up a national registry of clinical trials. That's also the issue behind a lawsuit filed this month by the New York state attorney general, Eliot L. Spitzer. The suit accused GlaxoSmithKline PLC of fraud for failing to publish all of the results from clinical trials it conducted on its antidepressant Paxil to determine whether the drug could help adolescents. One of the studies said it did, but others showed no impact. The former was publicized; the latter were not.

At the moment, the law is straightforward: Companies are obliged only to reveal the results of testing they have done on drugs before they are approved by the Food and Drug Administration. Any subsequent testing, which is usually done to determine if a drug approved for one disease might be effective against another disease -- or sometimes to compare the efficacy of different drugs -- can legally be kept confidential. Companies are not even forced to register the existence of all clinical trials on, a government Web site set up for that purpose, although they are theoretically meant to do so.

As we have argued before, it would be better, in principle, for more information about drugs to be available to the public in general and to doctors in particular; the American Medical Association recently said that its members feel the same way. But this is a change that has to be made at different levels. Concerned that they are being manipulated by drug companies, medical journal editors are planning to put together a proposal requiring companies to register detailed descriptions of their trials -- perhaps on an enhanced version of the site -- before agreeing to publish articles based on the final results. The National Institutes of Health, which runs the site, should assist them.

Forcing companies to publish results of all trials, as opposed to the fact of their existence, is more complicated, because a simple government regulation requiring publication of all results of all clinical trials might backfire and wind up discouraging companies from conducting any trials at all. For that reason, Congress -- not the New York courts -- needs to take up this issue again and look at incentives that might persuade companies to conduct more and better clinical trials, even if they aren't commercially advantageous. It should also consider establishing a routine, nationwide system of comparative drug testing, using university and other academic researchers. Any system that leads to the concealment or manipulation of research isn't serving doctors, patients or, in the end, even drug companies very well.