How likely a risk is has nothing to do with how bad it is. Where the likelihood is high and stake is large, few people do the risky thing: nobody willingly kicks bears. On the other hand, many people risk large stakes if they're unlikely to lose them: most people drive cars, even though fatal accidents happen. The true measure of a risk is its probability multiplied by the size of the stake at risk.
So what risks are acceptable? People take risks if they think that they'll gain more by taking the risk than they're likely to lose by taking it or because the cost of not taking the risk is too high. Most people are willing to drive to a grocery store despite the small risk of death or injury in order to save time (accept risk in expectation of coming out ahead) or because having the store deliver their groceries is expensive (accept risk because avoiding it is costly). This decision on what risks to take is called risk-benefit analysis.
Example: If you work outside during the summer, it's likely you'll be bitten by mosquitoes.
Example: It's possible that if you drive a car, you'll have an accident. Because of this, cars have seat belts to reduce your chance of injury if you are in an accident.
Example: The statistics of gas motion say that there's a very small chance that all the air in a room could rush to one corner, causing everybody in the room to suffocate. Nobody carries air tanks because they're afraid that this might happen, however.
Example: Normal adults don't worry about being devoured by monsters under their bed.