The commissioner of the Food and Drug Administration cautioned yesterday that fetal monitoring devices are being widely and routinely used without clear knowledge of the benefits or risk to infants.

Commissioner Donald Kennedy said that medical research done on fetal monitoring devices to date does not support flat assertions by physicians and manufacturers that the procedure is safe. Fetal monitoring, Kennedy told the Senate Health subcommittee, "is not known to be without risk."

Fetal monitoring involves two widely used technologies. One form known as ultrasound, applies electrodes to the mother's abdomen to detect internal events through echoes. The other, internal monitoring, threads wires and sensors throught the vagina and into the uterus, where one sensor is clipped to the fetal scalp and another is clipped on the uterus.

Both techniques provide doctors with information on the fetus' heartbeat and other vital signs.

Other witnesses at the hearing testified that the growing use of fetal monitoring devices has contributed to an almost doubled number of cesarean section deliveries in the United States between 1971 and 1976.

Dr. Albert D. Haverkamp said his studies show that the increased use of fetal monitoring devices has not improved the mortality rate for infants, their performance on physiological tests given immediately after birth, or their neurological condition.

At the same time, the cost of delivery increases to about $3,000 when cesarean section is performed from about $700 for a vaginal delivery, Haverkemp said. Haverkamp, an obstetrician-gynecologist in charge of high-risk believe we are in danger of over-selling our technology."

The subcommittee was also told that the growing practice of elective induction of labor - birth by appointment - involves a "much, much higher, significantly higher, unnecessarily higher" degree of infant mortality than is experienced in spontaneous births.

Dr. Roberto Caldeyro-Barcia, director of the Pan American Health Organization's Latin American Center of Perinatology and Human Development, said induced labor risks not only fetal death but also harm if the infant survives. He said that the birth canal may not be sufficiently softened and therefore is not "ripe" for birth, that the infant's head may not be as well protected, and the amount of "work" performances by the uterus may be greater.

As a result, Caldeyro-Barcia rejected such practices as prematurely rupturing the membranes containing the fetus. "The only induction of labor which is acceptable is the therapeutic induction," he said, in cases where continuing the pregnancy may increase harm to either the infant or its mother.

One or both types of fetal monitoring devices are widely and routinely employed by major hospitals across the country. According to data gathered by the Senate subcommittee, many hospitals - including some in the Washington metropolitan area - require obstetrical patients to be monitored. The subcommittee staff found that George Washington University Hospital, Georgetown University Hospital, Greater Southeast Hospital, Columbia Hospital for Women, Alexandria Hospital and Arlington Hospital monitor all or almost all obstetrical patients.

Experts, however, are divided on the potential risk. Some caution that ultrasound may cause damage to the infant that may not show up for years. Others say that rupturing the membranes to implant the sensors increases the risk of infection and may cause other problems.

"The time to find out," subcommittee Chairman Edward M. Kennedy (D-Mass.) said, "is before millions of children are exposed. Otherwise, we are playing an unjustifiable game of Russian roulette with the health of our children."

Dr. Richard Paul, of the University of Southern California department of obstetrics and gynecology, said he believes fetal monitoring has been "an important factor" in the halving of infant mortality at birth.

Ideally, Paul said, he would like to see fetal monitoring used in all births in all hospitals. But under questioning by Kennedy, Paul said some physicians do not have the required expertise. "It would be inappropriate for those people to use the tool, in the same way that we wouldn't let a 10-year-old drive a car," Paul said.