I still prefer the good old[tm] Labnotebook. No dependencies on (proprietary) software or a computer. I generate quite some digital data (blots, sequencing etc.) this is stored on the servers of my university. We looked into this in our lab, but discarded the idea pretty fast. Most solutions are either expensive and/or proprietary. Then a lot of these programs are not bought, but rented for a monthly or yearly fee, so once you are in you basically will not get out there anymore. Or it takes a lot of work.
I found one free (for up to three projects), this is Labguru.
There is an article about this on Bitesizebio, which has a few interesting comments. And there is another blog entry about this worth reading.
One of the problems with ELN is a legal one, since a labbook is a legal document providing evidence a what has been done and the results of it. They can be very important when it comes to patent legal battles.