Introduction:
Transitions between different tobacco products are frequent among tobacco users in Bangladesh; however, the reasons leading to such transitions and why they quit are not well researched. The aim of the study is to examine perceptions and reasons reported by tobacco users in Bangladesh to transition to other products or quit.

Methods:
Data from four waves (2009–2015) of the International Tobacco Control (ITC) Bangladesh Survey were used. Repeated data on perceptions and reasons for exclusive cigarette (n=520), bidi (n=130), and SLT users (n=308) to either start using other products or quit were analyzed with sampling weights. The percentages of responses across waves were used to calculate the pooled proportion data using a meta-analysis approach.

Results:
Common reasonsig for respondents switching to other tobacco products were influence of friends/family (73.8–86.0%), and curiosity (44.4–71.3%). The perceived calming effect of smoking cigarettes and bidis (43.2–56.9%), and the impression that bidis were less harmful (52.3%) and taste better (71.2%) were major reasons for exclusive SLT users to switch products. Health concerns (16.5– 62.7%) and disapproval from friends/family (29.8–56.4%) were generally the main reasons for quitting. For smoked tobacco users, doctor’s advice (41.6%), package warning labels (32.3%), and price (32.4%) seemed to be the major driving factors to quit.

Conclusions:
Results highlight that the reasons for switching between tobacco products and quitting include social factors (e.g. friends/family) and (mis) perceptions regarding the products. Tobacco control policy could emphasize cessation support, increased price and education campaigns as key policies to reduce overall tobacco use in Bangladesh. Data from four waves (2009–2015) of the International Tobacco Control (ITC) Bangladesh Survey were used. Repeated data on perceptions and reasons for exclusive cigarette (n=520), bidi (n=130), and SLT users (n=308) to either start using other products or quit were analyzed with sampling weights. The percentages of responses across waves were used to calculate the pooled proportion data using a meta-analysis approach.