TEHRAN, JUNE 27 -- As rescue efforts in the areas devastated by the earthquake begin to wind down, questions have arisen about exactly how many people died in the disaster, which Iranian officials describe as the worst earthquake in their country's modern history.

Wahid Dastjerdi, president of the Iranian Red Crescent Society, told a news conference today that approximately 40,000 people were confirmed killed in the quake. "These are not estimates, these are registered cases," Dastjerdi said.

The Red Crescent's numbers are lower than estimates put out by other arms of the government, which in recent days have tended to report the number dead at around 50,000. Other unofficial estimates, unsupported by evidence, have ranged as high as 70,000.

Dastjerdi did not elaborate about what he meant by 40,000 "registered cases," and it is not clear whether or how the government recorded the number of casualties, since many of the dead apparently were buried by family members without checking with authorities and many bodies remain under the rubble of collapsed villages.

Several visits to the worst-hit areas left no doubt that many thousands of people died in the quake. Two sizable towns, Rudbar and Manjil, were virtually obliterated. Reporters saw a number of villages near the towns that also were leveled.

But outside an approximately 20-mile radius of Rudbar and Manjil, considerably less destruction was visible. The relatively light damage in larger cities such as Rasht and Qazvin contrasted with early reports put out by Iranian officials that suggested large numbers of people had died there.

An early government report broadcast on Tehran radio suggested that Qazvin, a light industrial city of 300,000 about 75 miles northwest of Tehran, had suffered heavy losses. But the city emerged virtually unscathed, with hardly a broken pane of glass or crumbled wall seen.

An accurate estimate of the number of casualties would require knowing the exact populations of devasted Manjil and Rudbar, where it seems likely that a majority of the victims died. But according to diplomats here, the government has not conducted an accurate census since before the Islamic revolution of 1979.

Survivors and rescue workers have said at least half of the people living in Manjil and Rudbar last Thursday morning died when the earthquake struck, an estimate that appears conservative given the total destruction of the area and the relative dearth of survivors and refugees. What is less certain is the combined population of the towns before the earthquake struck.

Surviving residents and Iranian officials have said the combined population before the quake was 30,000 to 50,000, estimates that seem reasonable given the number of collapsed houses visible, but which even some Iranian officials acknowledge are only guesses.

From aerial tours and road trips of the barren region, it appears unlikely that more than another 20,000 people lived in villages or towns within 20 miles of Manjil and Rudbar. But no accurate population figures are available.

Visual evidence and the accounts of survivors and relief workers do not appear to call into serious doubt the Iranian Red Crescent Society's claim of 40,000 dead. But there appears to be less evidence to support the higher Iranian government and unofficial estimates. And there are few signs that a verifiable count of the dead will or can be made anytime soon.