The nightmares were the worst part, Craig Coleman, 32, will tell you. They intruded on his sleep for 10 months, always ending with the violent murder of his son. They gave him no peace until he quit working as a Baltimore County paramedic, a job he describes as "all I ever wanted to do with my life."

Coleman's voice still shakes when he recalls Christmas Eve 1983, when it all started -- the dreams, the headaches, the pain. For 30 minutes, after responding to an emergency mistakenly labeled "cardiac arrest," he had tried to persuade a man on the drug PCP to release his own 14-month-old son, held at knifepoint.

"I want my son to go to heaven," the man had yelled.

"I want you to bring your son to me," Coleman responded, "and I'll bring him to the hospital."

"I've killed the baby," the man replied. "There's nothing you can do."

The fact that Coleman probably had arrived after the child's death did not ease the trauma of that night, when he watched in vain while the baby's head was ripped from its body. The guilt set in, the haunting doubt that he "should have done something else or done it in a different way to save the baby's life."

Coleman sought psychiatric help soon after leaving the scene, and continued getting it for months, until he felt well again. Many other "first responders" -- those who have early contact with violent crimes, accidents and disasters -- don't get the help they need. But Jeffrey Mitchell, assistant professor of emergency health services at the University of Maryland's Baltimore County campus (UMBC), wants to change that.

Mitchell, a former firefighter and paramedic, earlier this month coordinated an international conference, "Stress and Behavioral Emergencies," at UMBC. The four-day symposium covered topics ranging from biofeedback to the psycho-emotional response to burns, but Mitchell's particular interest is post-traumatic stress disorder, a condition in which exposure to a serious emotional stress leads to problems such as recurrent dreams, a feeling of detachment from others and guilt.

Anyone who witnesses a highly disturbing event can be affected by post-traumatic stress disorder, he said, but its prevalence among emergency workers is only now being fully appreciated.

"We as a society have made the crazy assumption that emergency workers can pick up limbs and not be affected," Mitchell said. "I've lived the life of a paramedic. I've fought fires and seen the terrible reactions around me. I've seen the losses . . . I've listened to the family problems."

In Mitchell's 1984 study of 360 "first responders" in Maryland, Connecticut, Wisconsin and Illinois, 87 percent of fire service personnel complained of job-stress related symptoms such as sleep disturbance, irritability and headaches. Mitchell also cites a University of Texas study that placed the divorce rate among a particular group of paramedics at 65 percent.

Mitchell recommends that emergency-response facilities offer workers a program to deal with the psychological trauma involved in their jobs. Central to the program is a "critical incident stress debriefing," an individual or group meeting during which a traumatized emergency worker can talk with a counselor. Other components of the program include screening of emergency services applicants to hire individuals able to cope with high levels of stress, a training program to teach workers how to deal in a healthy way with their day-to-day emotions, periodic screenings to detect underlying psychological illness and family therapy.

When Mitchell first began to promote this idea, he met with some resistance.

"Ten years ago we sent out 600 invitations to a free conference on this," Mitchell said, "and four people showed up." Fire department administrators were, and many remain, wary that acknowledging the effects of psychological stress might alert workers to an easy route to disability payments.

"I see it particularly in large cities. They're afraid it will cost them a great deal if everyone jumps on the bandwagon," Mitchell said. "Sure, you'll get some of that, but most emergency people don't want to leave the job."

Despite some reluctance to initiate programs to deal with worker stress, Mitchell believes the concept is spreading rapidly. "Within this decade most major cities will have some phsychological support for emergency personnel," said Mitchell, who has organized support systems in half a dozen cities from Alexandria to Seattle.

One reason more programs are being implemented, Mitchell says, is the unusual number of recent tragedies -- the Air Florida crash, the Hyatt Regency disaster and others -- that have highlighted the need for these services. Lawsuits against fire and rescue departments also have started to result in out-of-court settlements favoring the individual stressed to the point of illness while on the job.

"People are beginning to realize that the program may be cheaper than no program at all," Mitchell said, pointing to an incident in Buffalo where 75 percent of an emergency unit's staff quit after a propane explosion killed five firefighters. "Unfortunately, we tend to learn out of tragedy."