Driving a car isn't dangerous? Only losing control of it is? Whereas with smoking the risk is not within the smoker's control? Using the word 'control' in both scenarios was probably not the best idea... I mean, if the driver loses control of the car, then by definition the risk of hurting someone else is not within the driver's control. When you choose to drive a car, I think you accept some inherent risk of danger to yourself and to people around you. This isn't even mentioning the 'invisible carcinogens' that are present in the exhaust fumes of cars.
It's really not difficult to come up with more examples. What about people who have the flu? Would you bar them from bars because they are spreading influenza virus? Estimates of death from second hand smoke are roughly 50,000 though it depends on how you causally link the two. Deaths due to influenza and pneumonia are nearly as high. Eating at a KFC while you've got the flu probably does more harm than smoking a cigarette in a KFC.
Shit, I'm not trying to downplay the dangers of smoking. There's a reason I'm not a smoker after all. I just think that it's really become a matter of morality now, and not one of consistency. Yes, there is an underlying harm to others aspect, but that aspect is present in so many human activities which are just, by and large, socially acceptable. Sure you can pick at the small differences between my examples and smoking, but do they really justify the huge disparity in legislative treatment? You can say that there's one extra step in causality between driving a car and crashing it, as though inhaling second hand smoke now is much more easily causally attributed to your death by cancer 30 years down the road. I just don't see the consistency.