At a more abstract level:
Keep a list of risks with all the possible risks that you can think of, be they big or small, probable or unbelievable. Refresh or at least double-check this list every few weeks. It can be as tangible as a “hard drive failure on the main server” or as intangible as “a competitor launches its product first”.
Then, for each risk, evaluate the impact and likelihood on some scale (this can be quite arbitrary). The real risk will be somewhat proportional to the product of both. IE, the high potential cost with a very low probability is not as bad as the average cost with a high probability. These numbers are just there to help you figure out the risk, don't take them seriously.
Next, for each risk, think about mitigating measures that you can take, be it countermeasures, insurance, whatever. Again, indicate the value of these (not on cash value!).
Only now can you really decide what (and if at all) to do with every risk. Just accepting risk may be an acceptable decision at this stage, but not before.
You might want to read Waltzing with Bears: Risk Management on Software Projects Tom DeMarco, Timothy Lister. Good time.
jdisk source share