After you've decided on the resolution of your finished image (or alternatively the pixel dimensions of your finished image), how do you maximize the image quality? There's one thing you can do which involves longer scanning times, but capturing more raw data for each pixel in the finished image.
The basic idea is scanning each pixel more than once and averaging the pixels. Each doubling of the number of pixels increases the effective number of useful bits of data by one. For instance, if you have a 10-bit scanner like the Nikon LS-30 and you read the CCD 4 times at each pixel position, you get effectively 12 bits of useful image data.
There are several ways of achieving multiple image samples. The first is single-pass multi-scanning. Some scanners are capable of reading each pixel position multiple times before advancing the scan head to a new position. The second technique is multi-pass multi-scanning, which most scanners are capable of (however, some can't accurately reposition each scan pass, so this may not always work well).
Another useful way of getting multiple image samples is to scan at a higher resolution and then average adjacent blocks of pixels. For instance, scanning at 2700 dpi and averaging every 2x2 block of pixels will result in a higher-quality 1350 dpi scan than just scanning at 1350 dpi. In this example Scanning at 1350 dpi throws away every other pixel and every other scan line, while scanning at 2700 dpi and setting Output | TIFF size reduction to "2" will result in averaging 2x2 blocks of pixels and increasing the number of effective bits of resolution by 2 bits.
Note that multi-scanning is the only way to increase the quality at the highest resolution, and that using Output | TIFF size reduction or Output | JPEG size reduction is a better way of producing quality scans at lower resolutions.