WASHINGTON (AP) _ The Enviromental Protection Agency wants to lower sharply its estimate of the cancer risk posed by dioxin, bringing it in line with estimates of other government agencies.

Even if EPA adopts the 17-fold reduction, dioxin still would rank as the most powerful carcinogen ever tested in animals and would be the most powerful of the ''probable'' human carcinogens.

An environmentalist critic of the proposed new policy said it would give ''carte blanche'' to build controversial municipal incinerators and would relax cleanup standards at Superfund toxic dump sites.

Dioxin is the short name for the most toxic of 75 related chlorine- containing compounds.

Dioxin causes chloracne, a skin disorder, and high concentrations are believed to cause birth defects, but there is conflicting evidence that it causes human cancer. A draft document being circulated to scientists for comment and made available by the agency on Wednesday said new studies of chemical accidents in Germany and West Virginia years ago seems to show elevated cancer incidence in the exposed groups, but assumptions about actual exposure are unsupported.

In estimating risk, government agencies use a so-called ''unit risk'' factor - in this case, the amount of dioxin that, ingested daily for 70 years, would increase a human being's chance of cancer by one in a million.

EPA has been using 0.006 picogram (trillionth of a gram) per kilogram (2.2 pounds) of body weight as the unit risk.

Now the agency wants to adopt a new estimate that it takes 17 times more dioxin - 0.1 picogram per kilogram - to increase the cancer risk by one in a million.

The draft said, ''Considering the strengths and weaknesses of each approach'' to estimating cancer-causing potential, ''0.1 picogram per kilogram per day has a reasonable scientific foundation and represents a rational science policy position.''

This brings EPA's unit risk into the range used by the Centers for Disease Control and only about 70 percent above that used by the Food and Drug Administration.

Many scientists believe that almost everybody carries about 10 parts per trillion of dioxin in body fat. The EPA draft said it is not clear this is true, but if it is true, it implies that people in general are taking in about 1.0 picogram per kilogram per day.

Under the old potency estimates, this would mean that ''background'' exposure to dioxin in the air, in food and by other routes would be producing at most about 300 cancers a year in the United States. The draft did not discuss what the new estimate would mean; but presumably the number of annual dioxin-caused cancer deaths would be less by a factor of 17.

About 483,000 Americans die of cancer each year.

The agency began a review of dioxin two years ago because EPA Administrator Lee M. Thomas wanted to try to reconcile EPA's much higher potency estimates with those of other agencies.

The new number is calculated as a ''plausible upper bound,'' meaning that the real risk of dioxin-caused cancer is unlikely to be higher than what is implied by the new estimate. ''The risk is almost certainly lower, and could be much lower,'' said Peter Preuss, head of the EPA dioxin review group.

Ellen Silbergeld, toxicologist for the Environmental Defense Fund, said EPA appeared to have picked a number out of the middle of a range of estimates arrived at by different means.

''The right answer is not the average of several right answers,'' she said. ''In science, there is only one right answer. ... This isn't science; it's accounting, and it doesn't take scientists to do it. All it takes is somebody with a calculator.''

If EPA uses the new estimate, ''It's carte blanche for incinerators,'' she said.

Many proposals for new municipal incinerators are being opposed by people who say they fear the dioxin that would be emitted in stack gases and put in the ash.

Dioxin is produced in tiny amounts in low-temperature combustion and is destroyed by high-temperature combustion. Burning certain plastics increases dioxin formation, and it is also found as an unwanted byproduct in the manufacture of certain chemicals, particularly some pesticides no longer made.