Significant errors may result when applying one-dimensional (1-D) or two-dimensional (2-D) interpretation methods to magnetotelluric (MT) data collected in three-dimensional (3-D) environments. Both depths and resistivities can be grossly incorrect if 1-D or 2-D methods are applied in a 3-D setting. We present examples of MT sounding curves generated using a 3-D modeling program (Madden and Park, 1982) and illustrate some interpretation pitfalls if 3-D effects are not considered. The 3-D effects discussed herein are attributed to a surface layer heterogeneity, and can be readily identified in MT data from a well-designed MT survey. The MT survey must include stations chosen to yield information about regional structure. Alternatively, carefull examination of a geologic map will help the intepreter estimate the regional effects present in the data.