Covid-19, the illness caused by the coronavirus, is arguably the leading cause of death in the United States right now. The virus has killed more than 1,800 Americans almost every day since April 7, and the official toll may be an undercount.

By comparison, heart disease typically kills 1,774 Americans a day, and cancer kills 1,641.

Yes, the coronavirus curves are plateauing. There are fewer hospital admissions in New York, the center of the epidemic, and fewer Covid-19 patients in I.C.U.s. The daily death toll is still grim, but no longer rising.

The epidemiological model often cited by the White House, which was produced by the University of Washington’s Institute for Health Metrics and Evaluation, originally predicted 100,000 to 240,000 deaths by midsummer. Now that figure is 60,000.

While this is encouraging news, it masks some significant concerns. The institute’s projection runs through Aug. 4, describing only the first wave of this epidemic. Without a vaccine, the virus is expected to circulate for years, and the death tally will rise over time.

The gains to date were achieved only by shutting down the country, a situation that cannot continue indefinitely. The White House’s “phased” plan for reopening will surely raise the death toll no matter how carefully it is executed. The best hope is that fatalities can be held to a minimum.

Reputable longer-term projections for how many Americans will die vary, but they are all grim. Various experts consulted by the Centers for Disease Control and Prevention in March predicted that the virus eventually could reach 48 percent to 65 percent of all Americans, with a fatality rate just under 1 percent, and would kill up to 1.7 million of them if nothing were done to stop the spread.

A model by researchers at Imperial College London cited by the president on March 30 predicted 2.2 million deaths in the United States by September under the same circumstances.

By comparison, about 420,000 Americans died in World War II.

The limited data from China are even more discouraging. Its epidemic has been halted — for the moment — and virtually everyone infected in its first wave has died or recovered.

China has officially reported about 83,000 cases and 4,632 deaths, which is a fatality rate of over 5 percent. The Trump administration has questioned the figures but has not produced more accurate ones.

Fatality rates depend heavily on how overwhelmed hospitals get and what percentage of cases are tested. China’s estimated death rate was 17 percent in the first week of January, when Wuhan was in chaos, according to a Center for Evidence-Based Medicine report, but only 0.7 percent by late February.

In this country, hospitals in several cities, including New York, came to the brink of chaos. Officials in both Wuhan and New Yorkhad to revise their death counts upward this week when they realized that many people had died at home of Covid-19, strokes, heart attacks or other causes, or because ambulances never came for them.

In fast-moving epidemics, far more victims pour into hospitals or die at home than doctors can test; at the same time, the mildly ill or asymptomatic never get tested. Those two factors distort the true fatality rate in opposite ways. If you don’t know how many people are infected, you don’t know how deadly a virus is.

Only when tens of thousands of antibody tests are done will we know how many silent carriers there may be in the United States. The C.D.C. has suggested it might be 25 percent of those who test positive. Researchers in Iceland said it might be double that.

China is also revising its own estimates. In February, a major study concluded that only 1 percent of cases in Wuhan were asymptomatic. New research says perhaps 60 percent were. Our knowledge gaps are still wide enough to make epidemiologists weep.

“All models are just models,” Dr. Anthony S. Fauci, science adviser to the White House coronavirus task force, has said. “When you get new data, you change them.”

There may be good news buried in this inconsistency: The virus may also be mutating to cause fewer symptoms. In the movies, viruses become more deadly. In reality, they usually become less so, because asymptomatic strains reach more hosts. Even the 1918 Spanish flu virus eventually faded into the seasonal H1N1 flu.

At the moment, however, we do not know exactly how transmissible or lethal the virus is. But refrigerated trucks parked outside hospitals tell us all we need to know: It is far worse than a bad flu season.

The lockdowns will end, but haltingly.