Experts are not sure why the numbers are rising. But one reason may be, paradoxically, greater access to health insurance.

An uninsured child who developed flu-like symptoms in the 1990s might have died from what was actually leukemia, but without medical care his death certificate said pneumonia, explained ACS's Brawley. With insurance, that child now is more likely to see a doctor and get correctly diagnosed.

Experts note that cancer incidence can be affected by such changes as how many people are screened for the disease, such as through mammography or prostate-specific antigen (PSA) tests. That may account for the recent increase in cases of thyroid and kidney cancers.

Incidence is therefore less accurate an indicator of progress against the disease than mortality. By this measure, too, progress has been halting at best, the report finds.

Death rates from cancer - the percentage of all deaths due to any form of this disease - have been declining since the early 1990s, by about 1.8 percent per year in men and 1.4 percent per year in women. But the decrease was also 1.8 percent per year from 2005 to 2009 for men, and 1.5 percent for women. So even as expensive new treatments were introduced, progress is not accelerating.

"We're all wondering why there hasn't been an acceleration in the rate of decrease" in mortality, said report co-author Edgar Simard, an epidemiologist at ACS.

One reason is that "cancer cells are wily foes," said Massachusetts Institute of Technology cancer biologist Robert Weinberg. "What is most frustrating is the ability of cancer cells that are under attack and being decimated by one therapy to invent a way of circumventing the therapy and finding a new way to thrive and proliferate."

Although the drop in smoking is responsible much of the decline in cancer death rates - 42 percent of adults smoked in the mid-1960s, compared with about 21 percent now - some of the credit goes to improved screening.

Colorectal screening can find and remove polyps before they become malignant, for instance. "So stage one never gets to stage two or three or four, and you have less mortality," said Dr Maurie Markman, senior vice-president for clinical affairs at Cancer Treatment Centers of America. Deaths from colorectal cancer fell 2.9 percent a year from 2000 to 2009 among men and 1.9 percent among women.

The 1.9 percent annual decline in deaths from breast cancer over the same period probably reflects some improvements in treatment. But "a huge driver of the decrease is what we call breast awareness," said Brawley. "It's not monthly breast exams" that do not decrease deaths "but women being aware of their body and asking, what is this change in my breast?"

The rise in liver cancer may be the result of a high prevalence of chronic hepatitis C infection resulting from drug use decades ago, the report says. Obesity may be raising the incidence of liver, pancreatic and uterine cancers.

"If the American public really wants fewer people to die from cancer, then there will need to be major changes in lifestyle," said MIT's Weinberg, such as declines in tobacco use and obesity. The biggest reductions in cancer mortality "will come from prevention rather than treatment."