If your 64gb fails, then 100% your files are gone, but if one of your 8gb fails or is lost, then 1/8 of your files are gone. I don’t think someone would carry 8 x 8gb cards anyways.

And if you backup all the time with those portable devices, then you can always format and use a smaller card.

Does anyone ever shoot 128gb worth of pics per event? That’s a lot of large RAWs and long HD videos.

Rule #1: Alex is right that small cards have relatively little reliability advantage, though his reasoning about probability is a bit off. Small cards will be a tiny bit better than the medium ones, but not significantly. But it’s best to avoid the newest technology (currently 64/128 GB). When the manufacturers are pushing the envelope, that’s when they’re still figuring out how to make things work every time. There’s nothing WRONG with small cards, but every time you swap cards you’re increasing the chances of some kind of mistake. So get a medium-sized card (by this year’s standards) and avoid swapping.

Rule #2: If you pay attention to anything on this blog post, pay attention to that one. And don’t just back up your cards during the shoot; back up your disks. Back them up off-site, too, so that a fire–or a burglar–can’t ruin everything.

Rule #3: Back pockets are probably a bad idea. But memory cards are WAY tougher than, for example, CDs. You don’t need to treat them like fine china. Just don’t do something that’s going to bend them–or snap them in half. And, like Alex says, be sure to keep track of which ones have been used!

Rule #4: This is the weakest of the recommendations. In the first place, copying a full load of images is time-consuming; in the second, looking at them is tedius and doesn’t really verify their integrity. For that reason, the manufacturers already test them before they go into the package. If you insist on testing cards yourself, it’s easy to do it with a program that will do bit-by-bit comparisons without human intervention. I’m sure you can find one somewhere on the Web (don’t ask me; I don’t use either Mac or Windows). But because of the way cards work, this kind of test isn’t going to reveal any but the grossest problems. It would be just as effective to take about 10 test shots and review them on-camera; if you don’t get error messages, the card is highly unlikely to fail at tomorrow’s wedding. (I can’t make any promises about six months from now, though. Cards can go bad unpredictably. See Rule #6.)

Rule #5: Nobody can argue.

And I’ll add rule #6: stuff happens. No matter how hard you try, sometimes things are going to go wrong. That was just as true in the film days as it is now. So when they do go wrong, relax and deal with it. Sometimes a recovery program can get stuff back. If not, well, you’ll have to make the best of it, just as if your clumsy lab assistant had managed to fog the most important roll you shot this week.