Personally, I don't do all that much "scientific" stuff as some people around here do. I always start with a raw file, convert that to 16-bit Wide Gamut RGB with my raw converter of choice (usually C1, with exceptions). Then I open it in PSCS2 and start cleaning up the sensor dust, bad CA if present, maybe do some light "Shadow/Highlight" (very slight, not a big fan of this tool, esp. since so many people are using it these days to such an extent that we get absurd looking images). Then I save this "raw" image for posterity =)
Now the 2nd step starts, print pre-processing. First I crop / pad to desired aspect ratio. Then I upsample in PS in one step using "Bicubic Smoother" to target size, 200dpi.
With this image, I once again do a once over looking for sensor turds that may have been overlooked before. Then I sharpen the image using USM - and that's where I go purely by gut feeling: I choose radius based on the upsampling factor that I've used, typically in the 1.0 - 3.0 range; then adjust Intensity to a level that "it just borderline hurts" - as in, it should look oversharpened on screen, but not too violently. Ink jet printers need more oversharpening than the LightJet.
The last step before printing is a Gamma adjustment of 1.05 ~ 1.15, depending on the printer. This is why it's neat that Aperture has this control built into the print panel
. The gamma value has to do with the fact that paper doesn't glow (yet, anyway) so typically an image on screen looks brighter than on paper, esp. given the lighting conditions where the image will be hanging; even a custom profile can't account for this because it doesn't know how bright it will be on the "target wall", all spectrometers use a constant light source.
Then I cross my fingers and print.
I have printed using this low tech approach pretty large pictures; 24x36 is commonplace for me, 60x40 has been done, too, from 1Ds, 1Ds2, and P45 cameras. The results are quite convincing, at least to me.
I don't like to get caught up in technical discussions about sharpening in 10 small steps vs. one big one; I don't discount the possibility that there's a difference, but I have never seen a compelling difference yet, plus I am an advocate of "proper viewing distance": Nobody is looking at my 60x40 pictures with a loupe, but more often from 5ft or more. At that distance, it doesn't really matter what sharpening technique I used, but rather what's the content of the image and how its execution / basic post processing was.