A laser works by exciting electrons in a very specific manner such that when the electron falls back into its ground state, it will send out a photon with a specified wavelength. But as one can show (done in http://www.uio.no/studier/emner/matnat/fys/FYS4110/h10/undervisningsmateriale/LectureNotes2010.pdf p 118, also see p 120) the photons that are sent out doesn't have the exact wavelength, but rather a finite width of wavelengths (or energies). So if you have a laser (say a Helium-Neon with wavelength 632.8 nm), you will get some photons with wavelength 632.5 nm etc. This could for example be a problem if you have a nonlinear material such as a crystal that you want to send your laser beam through, and the crystal behaves incorrectly to the photons with wrong wavelength.
I have only been doing calculations on a so-called three-level laser, which is extremely simplified, but in this case our assumptions were simply made, hoping that they are experimentally valid. In the real case, the line width is probably dependent of what material you use (such as some semiconductor), but I would guess this is hard to calculate.