The optical design of short-pulse CO2 laser-amplifier systems depends upon five considerations. First is the design for laser output, viz., the desired energy, power, frequency spectrum and pulse shape. These specifications determine the pressure, volume, and gain coefficient of the final amplifier stages, and the characteristics of the oscillator-switchout section. Second, constraints imposed by damage limits of the optical components, especially the output windows and the large beam relay and focus mirrors, fix the minimum laser aperture. Since the performance is limited by the maximum, rather than the average, optical power or energy, the gain should be as uniform as possible across the aperture. Third, parasitics must be suppressed. Although the limits on the small-signal gain which can be supported in CO2 amplifiers have increased with the discovery of nonlinear absorbers and optical blacks at 10-μm wavelength, parasitics always occur if the gain exceeds some limiting value. The value of the allowed small-signal gain influences the pressure and volume of the amplifiers through the interrelation with stored energy and pulse amplification. Fourth, the optical layout of the entire system must be designed to accommodate the considerations mentioned previously, as well as additional systems requirements for multipass energy extraction, efficiency, contrast ratio, beam expansion, aberrations, attenuation of target backscatter, and provision for alignment and focus. Finally, no matter how well the laser has been designed, the performance will be substandard if the optical components are not, or cannot, be manufactured to specifications. The LASL Helios and Antares lasers will be described to illustrate the influence of these considerations on the optical design of the present and future high-performance CO2 lasers, which are necessary for the development of laser fusion. Work performed under the auspices of the US Department of Energy.