We present an earthquake location scheme that uses the simplex algorithm in the solution. This method of direct iteration is conceptually simple, does not require derivative calculations, and avoids matrix inversions. The algorithm can be programmed on a small computer and its performance compares favorably with other methods of earthquake location. The method can be used for any velocity structure for which source to geophone times can be calculated. We use a horizontally layered velocity model, and consider both critically and noncritically refracted rays. It is easy to incorporate any time residual misfit norm into the location calculation. We generally use the L1 or L2 (least-squares) norm. The routine we present is used in two different microseismic monitoring systems at two potash mines in Saskatchewan, Canada.