Journal article: This paper investigates the relationship between an acute health shock, namely the first onset of an accident requiring medical care, and smoking, using rich panel data from a large French cohort of electricity board workers. To identify the causal effects of such shocks on smoking, we use a fixed-effect model. Results show a significant effect running from the shock to the number of cigarettes smoked with impact duration of 5 years after the occurrence of such shocks. Additionally, alcohol consumption is also reduced during 3 years, but the BMI is not impacted. Even though the decrease in the average number of cigarettes (i.e. 1,2 cigarettes less week) is quite low, this finding should be compared with average stopping or reducing attempts. Such attempts last, on average, 2,4 months that is less than 25 less than the decrease found here. Overall, our results show that health shocks seem to be a major determinant of tobacco consumption.