Just to revisit this a bit: "sharpness" is never absolute, so the issue is "acceptable sharpness" for the image and viewing conditions. In this age we magnify images (pixel-peep) to greater levels than we ever did with film, so being hyper-critical can find a but of blur that we never would have noticed before at normal viewing distance for a print. The old rule of thumb was to use a shutter speed faster than 1/Focal Length of the lens; or 1/50th for a 50mm, 1/90th for a 90mm lens, etc. But as I said above technique of holding and shooting makes a huge difference. When I said I used 1/15th hand-held without SR, I mean that it provides what I deem as acceptable sharpness for the way I use the pictures, but if you pixel-peep it won't be as sharp as a higher shutter speed. But if the picture still has the effect I want, so what?
Here are some illustrations of what I accept at different shutter speeds. This is a handy way to test your steadiness as well: shooting a mirror hand-held with the ISO and lens set to give you the speed you want to try. With digital it is easy to practice and work at improving. For this I used a Leica M9 (which has no SR) with a 90mm f4 lens; so the rule of thumb would be to stay above 1/90 sec speed. But here's a snapshot at 1/15th sec handheld, free-standing:
No, it isn't perfectly sharp, but is acceptable for normal viewing. For those who pixel-peep, here's a 100% crop to show the amount of blur, followed by crops taken at 1/25th and 1/60th second for comparison.
1/15th second, 90mm
1/25th second, 90mm
1/60th second 90mm
I didn't see a need to go faster than 1/60, as the differences would be slight.
So practice, and find out how much your technique can improve.
BTW - these were all shot at f4 - wide open, so there can be a focus variation with body sway as well. The M9 has no way to tell the actual fstop used, but approximates based on brightness compared to external sensor. I varied ISO to change shutter speeds.