This has implications for the true extent of the impact of smoking on health, say the researchers, who point out that the current statistical estimates of the death toll from smoking are potentially flawed.
They looked at just over 2,000 death certificates and 236 post mortem reports, issued at a large London teaching hospital between 2003 and 2009, to see what cause of death doctors had cited.
Doctors have been allowed to cite smoking and alcohol as a direct or underlying cause of death without the need to refer the case to a coroner since 1992.
Smoking was identified as the cause of death in only two certificates (0.1% of the total) and included in part II of the death certificate, which outlines other contributory conditions, in only 10 cases (0.5% of the total).
The two cases in which smoking was cited were lung cancer and chronic obstructive pulmonary disease (COPD). Yet 279 deaths included these diagnoses, and in most cases the deceased was a current (over 45%) or former (over 23%) smoker. It is well known that smoking is the primary cause of both lung cancer and COPD.
In all, 407 deaths were caused by conditions in which smoking is thought to have a substantial role. Yet smoking was cited as the cause of death in only two of these certificates and as a contributory factor in six.
The post mortem reports were no better: not a single case cited smoking as causing or contributing to death, which the authors describe as "surprising."
Yet doctors willingly cited alcohol as a direct or contributory cause of death. This was cited in over half (57.4%) of the 54 death certificates, which included diagnoses linked to alcohol use.
"Death certification is an important source of mortality data and directly captures 99.79% of all deaths in the UK," say the authors, who point out that the doctors in this study are not unique in their reluctance to cite smoking as a cause of death.
"There are many reasons why smoking is not cited as a [cause of death] by doctors in the UK," they write. "The first and frequently debated reason relates to doctors' desire not to cause relatives distress by stigmatising the deceased and their smoking habit."
They continue: "While the results of this study would support this assumption, it is interesting that the same clinicians frequently cited alcohol use as an underlying cause of death."
This may be because alcohol use is generally more accepted culturally, suggest the authors, adding that the stigma associated with smoking is well documented, and may be worsening as a result of the recent legislation, banning smoking in public places.
"Given the overwhelming evidence showing a causal link between smoking and certain terminal conditions, more effort should be made to record smoking on the death certificate. It is clear that the current arrangements fail to achieve this," they conclude.