We undertake a reevaluation of the surface‐photovoltage diffusion‐length experiment, numerically solving the governing equations without any prior assumptions. We find that, although the technique is applicable for crystalline semiconductors, it does not measure, in general, a simple diffusion lengthLD=(kT&mgr;&tgr;)1/2for amorphous materials. It is demonstrated that this failure is attributable to the localized gap states inherent in amorphous semiconductors. The presence of localized gap states results in drift and diffusion terms of comparable magnitude and, hence, results in the inability to simply measure the (&mgr;&tgr;)1/2product. We also demonstrate that using thicker samples does not, in general, alleviate this problem.