Lung cancer risk in smokers of different types of cigarette was compared ba
sed on evidence from 54 epidemiological studies, each of over 100 lung canc
er cases. Random effects meta-analyses estimated the relative risk of lung
cancer in filter and plain cigarette smokers ( or with most and least filte
r use), in lower and higher tar smokers, in ever handrolled and manufacture
d only smokers, and in ever black tobacco and blond tobacco only smokers. F
rom 43 gender-specific estimates, the risk was estimated to be 36% (95% con
fidence interval 27% to 44%) lower in filter than in plain cigarette smoker
s. The reduction was evident in both genders and in different continents. I
t was greater, 50% (33% to 63%, n = 11 estimates), for squamous-cell carcin
oma but was also evident, 20% (-6% to 39%, n = 8), for adenocarcinoma. The
risk was 23% (12% to 32%, n = 22) lower for lower tar than higher tar smoke
rs. This reduction, seen in both genders, equates to a 2% to 3% risk reduct
ion per milligram tar per cigarette. The risk was increased by 42% (21% to
66%, n = 15) for handrolled cigarette smoking and 75% (47% to 109%, n = 12)
for black tobacco cigarette smoking. These estimates did not depend critic
ally on results from specific studies or the limited confounding adjustment
in some studies. Various problems with the epidemiological evidence are di
scussed, including the difficulty of obtaining compatible exposure indices
from each study, the inadequate reporting of data from the largest studies,
the likelihood of misreporting of cigarette type, and the inadequate contr
ol for nonsmoking confounding variables. Difficulties also arise in adjusti
ng appropriately for aspects of smoking behavior that not only may differ b
etween those choosing to switch or not to switch to lower tar but also may
change as a result of switching. The overall data clearly show, however, th
at lung cancer risk is affected by the type of cigarette smoked. The sugges
tion that the switch to low tar cigarettes has led to an increase in the sm
oker/nonsmoker lung cancer relative risk and in the relative frequency of a
denocarcinoma versus squamous carcinoma is shown to be weakly based. Althou
gh more data are needed relating to modern very-low-tar cigarettes, the evi
dence is consistent with tar reduction and the switch to filter cigarettes
having reduced lung cancer risk. Indeed, the meta-analysis estimates may se
riously understate the risk reductions associated with lifetime smoking of
low-tar filter cigarettes.