We examine the effects of smoking on productivity among agricultural workers in riverine islands (locally known as chars) of northern Bangladesh, where prevalence of tobacco consumption is around 80% compared to 35.3% nationally. There is a high correlation between physically-demanding occupations and smoking, wherein farmers and day laborers are among those most likely to smoke. This means the opportunity cost of smoking is potentially very high for people employed in the labor-intensive agriculture sector. We use primary data from the Bangladesh Chars Tobacco Assessment Project 2018 survey for our empirical analyses. The effects of smoking on agricultural productivity is modelled using a standard Cobb-Douglas production function, with an additional parameter to capture the effect of the primary farmer’s smoking status on productivity. We estimate the effects using a two-stage non-linear least squares (NL2S) model through its impact on effective family labor. Our results show that smoking by the primary farmer reduces productivity of effective family labor input by 60-62%. Public policy objectives to improve labor productivity in the riverine islands of Bangladesh should actively target smoking behavior of agricultural households.