Unfortunately, one cannot simply double the number of scans in order to double the S/N, because the signal and the noise do not build up at the same rate. Instead, the signal increases as N^1/2, where N is the number of scans. In other words, to double the signal, you need to quadruple the scans; to quadruple the signal, the number of scans must be increased by a factor of sixteen. Since the length of the experiment is directly proportional to the number of scans, this approach to enhancing the signal can quickly become extremely time-prohibitive. When running a ¹H spectrum, the increase on going from, say, 16 scans (<2 minute experiment time) to 64 (about 5 minutes) to double S/N is not a particular hardship, but the same cannot be said for a typical ¹³C run (1024 scans with a 5 second pulse delay), which might increase from 1.5 to 6 hours.

The number of scans is one of the most commonly changed experimental parameters. Sample variables, such as concentration or viscosity, can make selection of a larger-than-standard number of scans advisable. In general, nuclides that are relatively rare (such as ¹³C at 1.1%, ¹⁵N at 0.37%, or ²⁹Si at 4.7%) require much longer experiment times than those that are naturally abundant (such as ¹H at 99.99%, and ³¹P and ¹⁹F at 100%) at the same concentration. Furthermore, those with lower Larmor frequencies (like ¹⁵N, ³⁵Cl, or ²³Mg) are inherently less sensitive than those which resonate higher (like ¹H, ³H, or ¹⁹F), and there are other factors that can affect the sensitivity as well. To learn the fundamental NMR properties of a particular nuclide, consult an NMR Periodic Table.