The choice of a chemical equation solver for a three-dimensional atmospheric model ultimately comes down to a tradeoff between speed and confidence in the accuracy of the code under a variety of conditions. Early codes that treated chemical kinetics in 3-D included family chemistry schemes, implicit or analytical schemes with a limited number of interations, and hybrid schemes that combined family techniques and implicit/analytical techniques. These codes were often compared with Gear's code in a box model for accuracy under certain conditions, then applied to a variety of conditions in the atmosphere. Gear's code, itself, was not applicable to 3-D atmospheric studies prior to 1993, since the original code required too many computations in matrix calculations and too much overhead iterating in one grid cell at a time. Since then, Gear's code has been improved in terms of its speed and applied in 3-D on regional and global scales, reducing the accuracy problem that plagued many early codes.  At the same time, gas chemistry is no longer the limiting factor in many models; hence, the use of an intense integrator is not so problematic as it once was.  In the 1990s, other accurate integrators aside from Gear's method were also improved or developed, but few have been applied in practice to 3-D.  Several hybrid schemes have also been improved and used in 3-D, but many such schemes have not undergone stringent accuracy tests under a variety of conditions needed to maintain confidence in their calculations, or the errors are recognized and accepted. This talk will discuss some of the chemical integrators available in atmospheric models and the application of a Gear-type code modified in terms of its speed and used to solve regional and global 3-D  atmospheric chemistry problems for the past seven years.  The performance of Gear-type code on scalar and vector computers will also be discussed.   Other solver techniques, including the approaches of Sandu et al and Hertel et al will be outlined.