Tests on animals are a much more accurate indicator of real human hazards -- at least as far as reproduction is concerned -- than scientists have generally believed, according to Dr. Nate Karch, who just completed a study for the Council on Environmental Quality.

Karch reviewed animal studies of 21 chemicals that are known from other evidence to be hazards to human reproduction. He found a strong similarity not only in the way the chemicals damage reproduction, but also in the doses that cause the damage.

The study even found that for 20 of the 21 substances, humans are affected by smaller doses than the animals, when the doses are corrected for the animals' size. The substances tested ranged from alcohol to thalidomide to diethylstibestrol (DES).

The study called for more research, saying evidence increasingly links some chemicals to all sorts of reproductive problems -- including infertility, spontaneous abortion and birth defects -- while fewer than 5 percent of the 55,000 chemical substances in commercial production in the U.S. have been tested for their effects on reproduction.

Furthermore, the study said sterility and spontaneous abortions may be much larger problems than birth defects: Estimates of the number of attempts at conception that fail range up to 80 percent, the study said, and more than 15 percent of pregnancies end in spointaneous abortion.

Karch, who is with Clement Associates, the research organization that did the report for CEQ, said it was still unclear how many of those failures are caused by chemical interference, but he said the number in "substantial."

"We think it is a big problem because there are a lot of chemicals that have been found to have bad reproductive effects," he said, "and in this study we have shown a surprising concordance between the negative efffects on animals and those on humans."