Chemjobber has been musing on the carnage a major disaster can have on lab equipment and research, and perhaps a few ways to both evacuate and ensure that your equipment is safe. But for every hurricane/flood/earthquake scenario there are dozens or hundreds of smaller disasters caused by simple accidents. I’m thinking of the minor floods that come when the tubing of a reflux condenser slips off in the middle of the night, or the solvent still fire that takes out an entire hood. Good laboratory practice will cut down on these incidences, but even if you do everything right there’s no guarantee that the undergrad in the lab above is as thorough.
Since there’s no warning before these accidents, the only solution is to reduce your vulnerability. Start small, by backing up your NMR data and scanning copies of your lab books. Cloud storage is godsend, and several prominent companies offer gigabytes of free storage . For a true backup the data should also be burned to DVDs periodically, with the discs stored off campus.
Key intermediates should be stored in 4-20mL vials and kept in a -20°C freezer, preferably in a second room (ie. one with no solvent stills or active research). If the weight of the vials when empty is written on the side in permanent marker the amount of material left can be quickly calculated, and no sample is lost when the vial is near empty. For a small number of vials I prefer to use resealable bags, but for thirty or more it’s better to store them in pipette boxes or the original package, in order of synthesis.
Keeping with the possibility of small floods, computers and other electronics should be stored at least 5cm (2 in) off the ground. For the truly paranoid their tops can also be covered, to prevent damage by sprinkler systems. Keep in mind though, in the event of a full-blown lab fire water from fire hoses will almost certainly destroy all electronic equipment, so good insurance is a better option. A more likely problem is a power surge or minor blackout, for which an uninterruptible power supply is a cheap solution.
As you work, tidy up loose ends. Trusty experimental procedures should be copied out of the lab book and written to publication standards, reducing the importance of your lab book. Confirmatory HRMS, HPLC and NMR analysis should be done prior to long term storage, and the results again added to a draft publication’s experimental section. The closer a project gets to publication the less important the early materials are, provided everything has been properly documented.
 Google, Microsoft and Amazon are all in this game. We use dropbox, for the Linux support.