Scientists are calling for a review of cigarette tests, and say the tobacco industry uses a machine which gives misleadingly low results.
Publicly available documents were scrutinised by researchers
A Canadian team, which reviewed British American Tobacco documents, said nicotine and tar levels measured are far below what smokers take in.
In the Lancet, the team suggests BAT maximised these discrepancies as part of marketing campaigns.
But BAT said it had publicised the disparities, and wanted a better test.
It said the industry and health bodies had been aware of problems with the International Smoking Organization (ISO) test - where a cigarette is inserted into a machine to measure the residues of tar and nicotine - for over 30 years.
Researchers from the Division of Health Studies at the University of Waterloo, Ontario, and Physicians for a Smoke-Free Canada looked at publicly available documents relating to smoking behaviour from the tobacco industry.
These conclude that, unlike the testing equipment, human smokers adapt to low-nicotine and low-tar cigarettes by taking deeper and more frequent puffs.
The researchers say the company exploited this discrepancy by developing so-called elastic cigarettes.
For example, on one brand that was tested, when humans inhaled just over a quarter more smoke, compared to the machine, it delivered almost twice as much nicotine and more than double the amount of tar.
The paper says the failure to disclose this product strategy and its health implications is a breach of public trust
Writing in the Lancet, the team led by Professor David Hammond said: "The discrepancy between ISO yields and the nicotine and tar delivered to human smokers is not simply an historical accident, but exists by careful design."
Team members quote from an undated presentation given by Colin Grieg, a senior BAT researcher, to colleagues in which he said the idea was: "... to produce a cigarette which can be machine smoked at a certain tar band, but which, in human hands, can exceed the tar banding."
The Canadian researchers say these, and other documents, show "a deliberate strategy whereby BAT and ITL [a subsidiary] designed products that would fool their consumers and regulators into thinking these products were safer or less hazardous when they were not."
They call for new, more accurate, measurements to be brought in to replace the ISO measurements.
Jean King, director of tobacco control at Cancer Research UK, said: "Consumers must know what they are consuming, which means that we need better testing methods for cigarettes.
"Tobacco companies must be prevented from manipulating their products and marketing them to mislead consumers into thinking cigarettes are less hazardous than they really are."
She added: "It is unacceptable that cigarettes are currently less rigorously regulated than food products or nicotine replacement therapy."
However, a spokeswoman for BAT challenged the Lancet paper.
She said the industry had publicly acknowledged the disparity between the ISO test and smokers' intakes of nicotine and tar in 1967.
She added: "This machine standard has been laid down by governments, that it must be used.
"It's not our regulations that say these readings must be on the packets. It's governments and the EU and medical authorities who say so."
"Scientists from BAT and scientists from the World Health Organization and the broader public health community, including one of the authors of this paper, have worked together on a scientific panel considering a new ISO standard."
She also said smokers who switched to low, or ultra-low tar cigarettes did change their smoking behaviour - puffing for longer or smoking further down the cigarette - but did not smoke more cigarettes.