You absolutely should include the base exposure with the overexposed and underexposed images when photomerging. rgs, when you say you select which exposures to photomerge, are you choosing all the ones that fall in-between your most overexposed vs underexposed? If not, then you're not photomerging 1EV exposures anyway (and is self defeating in trying to get the most tonal information). Ideally, you should photomerge all your exposures since you need all of them to fill all stops of light (including the various tones that aren't visible in your 8bit preview). Well if the argument for 1EV bracketing is to reduce noise, that doesn't make sense to me. Noise appears if you try to adjust your curves in the shadow areas ( and has the least amount of data anyway). One of the reasons expose to the right is a recommendation for digital exposures is that there's more tonal data in bright areas vs dark. If you're missing data in the highlights or midtones, it would show up more as banding or clipping rather then noise. I've only tried 7 bracketed exposures a few times....but now I'm wondering if I should try running an experiment this weekend to compare identical brackets with 1EV or 3 to see if there's differences in clipping with the bright areas. The main argument for 1EV brackets (making sure you merge all your photos 1EV), is that there may be more tonal information recorded. But the overall dynamic range is reduced because you're now missing 2 or 3 stops of highlights and 2 or 3 stops of shadows. You got me wondering, since I don't think there's as drastic a difference in noise or banding when merging all your RAW exposures...will just have to go out and shoot to see for myself