Ecologists have been studying the relationship between species richness and area for about a century. As area increases, more species are typically observed. Many mathematical functions have been proposed to describe the pattern of increase. Numerous researchers have assumed that the relationship is a power function despite the fact that there are many possible alternatives. There has been limited work in evaluating which species-area functions are most appropriate for field data. This study examines which of a variety of functions best describe how Sonoran Desert plant species richness of remnant habitat patches in the Phoenix metropolitan area vary with sampled area and the area of entire patches. No single species-area function was adequate for describing all empirical datasets. Sample curves of woody species were most frequently best described by the sigmoid logistic, Hill, and Lomolino functions, whereas herbaceous datasets were best fit by the sigmoid logistic or convex rational functions. A curve depicting the relationship between patch-level woody species richness and patch area was best fit by the convex exponential function. The power function provided the best fit for only one case. This study demonstrates the utility of testing alternative functions for statistical fit rather than assuming that any particular equation adequately describes the species-area relationship.