There is a certain small group of serial killers who are cruel, ritualistic and devoid of conscience, whose crimes have a strong sexual component and often involve rape or torture. They are outwardly normal and often very intelligent. They are difficult to apprehend and convict because they cover their tracks and improve their methods with each calculated crime. Because of that difficulty, they sometimes are able to plead guilty to lesser offenses and thus avoid maximum sentences. When they are released, it is almost certain they will kill again. Those of us who study them call them "lethal predators."

These people are not legally insane; they understand their misbehavior and can choose when to act upon their urges. Rather, they are mentally abnormal--"a rare blend of psychopathy combined with predatory sexual violence," in the words of Canadian psychologist Robert Hare.

FBI profilers Alan C. Brantley and Mary Ellen O'Toole estimate that there are approximately 100 such criminals in American prisons who are eligible to be paroled or released in less than a decade. And that should not be allowed to happen. Lethal predators serving less than a life sentence should be confined beyond the terms imposed in criminal court. This is not double jeopardy (two punishments for the same offense). It is a constitutional way to save innocent lives.

State by state, legislators can pass laws to accomplish this goal. This has been feasible since 1997, when the U.S. Supreme Court declared that Kansas's Sexually Violent Predator Act was constitutional. Kansas was one of several states that passed laws allowing the involuntary hospitalization of certain sexual predators after they complete their criminal sentences. Various state courts found those laws unconstitutional. Double jeopardy was one reason. Another was the fact that most of these criminals could not be effectively treated, and, in the words of an earlier Supreme Court ruling, "Absent treatment, confinement is imprisonment." This is ironic: If they can't be treated, why should these criminals be sent back into society?

But in the case of Kansas v. Hendricks, the Supreme Court said states could "quarantine" criminals who have a "mental abnormality" that results in predation, even though they had already served their sentences. Such hospitalization must be humane and nonpunitive, the court said; treatment may be offered, but a cure need not be expected. The court drew a parallel between such a predator and a person whose contagious disease poses a grave risk to others; either should be isolated, curable or not.

Although such civil commitment after confinement is now constitutional, only a handful of states have enacted legislation that would make that possible. More states need to do so.

I never thought I'd be a partisan in this debate. After many years in state and federal mental health administration, since 1981 I have been studying and treating the victims of violence, and avoiding the perpetrators. But then the case of Donald Gene Miller landed on my doorstep.

Miller was a 23-year-old criminal justice student and youth minister in East Lansing, Mich., when his fiancee disappeared on New Year's Day 1977. Two years later, Miller was convicted of raping and nearly strangling a 14-year-old girl, then stabbing her 13-year-old brother. Local prosecutors were sure Miller was responsible for the disappearances of his fiancee and at least three other women, but they lacked admissible evidence.

So the prosecutors offered a deal: Miller would lead authorities to the bodies of his victims and he would plead guilty to four counts of manslaughter. In exchange, his sentence for those killings would run concurrently with that for rape and attempted murder of the teenage girl--30 to 50 years.

Miller accepted the bargain, the bodies were identified, and he went to prison. Due to statutory sentencing guidelines, including mandatory time off for good behavior, Miller was to be freed in February 1999.

In 1997, my colleagues in the Michigan Victim Alliance asked me to help: Can't we do something, they said, to keep this serial killer from coming back to East Lansing?

As former mental health director for the state of Michigan, I was able to assemble a group of detectives, judges, legislators, prosecutors, prison psychologists and victim advocates. For two years, we debated Miller's diagnosis, prognosis and dangerousness. We interviewed experts who had interviewed him, including Earl James, a state police homicide detective and author of the book "Catching Serial Killers." "Clearly," James said, "he fit the profile of the intelligent, organized predator who escalates from fantasy to torture, murder and necrophilia." Nothing concentrates the mind like knowing a man like this is coming home--and his home is in your neighborhood.

Could Miller be confined in a mental hospital? Not under Michigan law--he wasn't insane; he was evil. Could Miller be tried in federal court for civil rights violations (crimes against women)? Not with existing evidence and binding non-disclosure agreements.

Ultimately, we discovered that Miller could be charged with a prison felony. Authorities sometime earlier had found a garrote in his cell and had given him a slap on the wrist--docking two years of his earned "good time." Our committee got prosecutors to bring this offense to criminal court. Miller was convicted. At sentencing, we were able to introduce evidence of his sadistic psychopathy because Michigan has a habitual offender statute and this was his third felony conviction. The court tacked on 20 to 40 more years.

Obviously, we were lucky. Other communities across America, facing the releases of their Donald Millers, are very much at risk.

The National Center for the Analysis of Violent Crime, a unit of the FBI, is now working with members of our Michigan team to reduce that risk. We are summarizing the scientific findings on the dangerousness and recidivism of predators, developing protocols for use in identifying incarcerated lethal predators, and helping state legislators frame appropriate laws.

Although the Supreme Court found the Kansas statute constitutional, it is not an ideal model, since it is broad enough to allow indefinite hospitalization of thousands of sexual deviants, including relatively harmless habitual exhibitionists. Laws that are too broad could be too costly, overwhelm the mental health system, and remove incentives for treatment where treatment is possible.

These are some of the valid reasons cited by the American Psychiatric Association and the American Civil Liberties Union for their objection to the Kansas v. Hendricks precedent.

But it is possible for states to write effective statutes that affect only a small group of genuinely dangerous killers. They can define lethal predators and require that such prisoners be evaluated for post-release civil commitment.

The inmate obviously would have to be lethal--he must have killed--and he must be a predator--that is, he must have left evidence of repetitious predatory acts, whether or not they resulted in criminal convictions. One essential feature of predation is its motivation: The perpetrator gets a sexual pleasure--usually involving arousal--while causing pain to the victim. Another feature is the crime's ritualistic quality.

A lethal predator also must demonstrate mental abnormality, usually a combination of sexual sadism and psychopathy. A psychologist or psychiatrist can evaluate sexual sadism to a high degree of reliability using standard tests, clinical interviews and evaluation of corroborated evidence. Psychopathy can also be determined. (This is not to be confused with psychosis, in which a person is out of touch with reality; psychotic criminals were judged mentally ill and confined long before Kansas v. Hendricks.) A psychopath understands right from wrong but has no conscience. Canadian psychologist Hare developed a commonly used method for establishing this condition, which consists of 20 items that trained assessors can rate according to a carefully constructed formulation.

At the civil commitment trial, a judge or jury would hear the testimony of police investigators and psychiatrists or psychiatrists, and then decide whether the inmate fit the statute. The standard of proof should be "clear and convincing" rather than "beyond reasonable doubt," since this is a civil, not a criminal determination. Predators should be confined in a secure setting, but separate from nonpredatory mentally ill patients. Each individual's case could be reviewed at set intervals.

By limiting the scope of new laws to the most dangerous class of predators, objections based upon cost and mental health policy will be avoided. Lives will be saved. And our state legislators will be spared the embarrassment of releasing lethal predators back home--to where we live.

Frank Ochberg, a psychiatrist, is an adjunct professor of criminal justice and journalism and a clinical professor of psychiatry at Michigan State University.