A measurement system using a CCD camera was developed for the objective evaluation of fabric wrinkle recovery. The shape of a fabric wrinkle was characterized by black and white color information or power spectra through the Fast Fourier Transformation. Artificial neural networks with the input patterns designed by two types were used to evaluate the fabric wrinkle objectively. These neural networks were constructed on the basis of Generalized Delta Rule. On the other hand, five human experts determined the subjective rank of fabric wrinkle recovery by AATCC TM/128. Wrinkle recovery rank evaluated by neural networks had a little relation to rank generated by human exports. The effects of fabric color on the rank were also investigated.