You guys mentioned that the colors are there- but you missed my point- that we need instruments to perceive them. To the naked eye, the objects are not bright, are not color saturated. Even if we were close to them, they would appear dim, and the colors would be fleeting, at best. This is well-known and not really debatable. Colors represent energy levels, and that is the way we perceive the differences in those energy levels. For deep-sky astronomical objects, we generally need large aperture or time exposures to bring the flux level up to the point where we can percieve the colors and see the dim, outer regions of deep-space objects. It's analagous to looking at an x-ray or image of any other spectral region that is beyond our unassisted vision, either because of wavelength or intensity threshold. Galaxies do not look like their photos, in real-time, naked-eye views, regardless of viewing distance. Ever. The images are always representational and more analagous to art (or even a sort of intensity map of flux levels) than a "xerox copy" or a daylight snapshot of an everyday scene. Joe mentioned that some people may perceive tints differently and this has also been documented. We've discussed in this forum before how some humans have additional color receptors in their eyes that allow them to see colors completely hidden from most of us. One English woman with the condition would laugh at her friends who thought they were wearing color-coordinated outfits, but to her eyes, they clearly were not! It took her many years before her wondrous ability was recognized- she sees with a richer color pallete than most of us. There is no doubt that she would see galaxies as being tinted differently than we do. So, I maintain a "yes and no" attitude on what is "correct" color in deep-sky objects. Certainly a lot of what we think of as correct is based on learned behavior. What we are used to seeing, what NASA and NOAO has spoon-fed us over the decades. There is an established paradigm for these objects and we frame our judgements based on that model. Is it correct in an absolute sense? No, for the reasons outlined above. Is it correct statistically? Is there a portion of the bell curve of color processing that we can assign as having more validity than the extremes? That answer is a *qualified *Yes.