Because Smart Objects are composed of mathematical formulas rather than pixels, they may be sized and re-sized both larger and smaller with no loss of quality. PS CS5 offers a number of ways of converting a pixel-based image into a Smart Object
One very convenient method for someone who does portraits, and who would like the background of the subject to be blurred, is to go to the Filter menu and select "Convert for Smart Filters." CS5 will convert the image into a Smart Object on a New Layer, at which time the image becomes fully editable.
If one then applies a Gaussian Blur to the entire image, the result will be not only overall blurring but also the automatic placement of a mask. If one then paints over the model with black paint while the mask is active, one is left with a portrait in which the model appears to have been shot with a telephoto lens -- she is in focus but everything around here is not.
One gets this result even when the photo was taken with a 35-50mm lens from only a few feet away!!!
Bonus: Because everything but the model is deeply blurred, there may be no need to remove "dust spots" or otherwise edit the surrounding background -- a significat time-saver.
In short, this method gives an improved result AND saves time.
My hat is off to Bob for providing some useful links for learning about Smart Objects.