Euthanizing animals was a huge part of my job. Labs don't usually use the reliable, expensive chemicals that vets use to put pets "to sleep," generally because the chemicals can affect postmortem laboratory results. So, in order to render animals into analyzable samples, scientists get somewhat inventive. One method for mice, referred to by the hair-raising euphemism "cervical dislocation," involves breaking the animal's neck, sometimes with your bare hands. For very young mice, you cut off the heads with scissors, or the aforementioned guillotines. Science!
Still, the most common method of euthanasia for many species, due to its cheapness, ease (for the human performing it), and speed, is suffocation with carbon dioxide. A lot of studies have examined whether the animals suffer during this -- the results are always inconclusive, making critics of the studies question the logic of endlessly killing animals with gas just to check if we can kill animals with gas.
Labs don't report how many animals they kill, so no accurate numbers are available. But based on the number of mice I put down on an average day (about 150 to 200), and the number of comparable labs in the country, I would estimate that U.S. labs kill upwards of 16 million mice per year (the Humane Society estimates 25 million). For scale, that's about 40 football fields packed tight with a layer of mouse corpses.
Or enough to lower the global happiness index a couple of points, if you prefer.