Currently there are no effective standardized tests of face perception without memory demands. This leaves us unable to determine if face recognition impairments are caused by perceptual or memory problems. Experiments with normal subjects also require assessment of face perception abilities. To address these issues, we have created a face perception task in which subjects must order simultaneously presented faces based on their similarity to a target face. The experiment consists of twelve sets; in each set a target face is presented along with six morphed test faces that each resembles the target face to a different degree. The test faces were created by morphing the target face with another face; different proportions were used to achieve different levels of resemblance. On some sets, we created each test face by morphing the target face with a different face, thus minimizing subject's ability to see continuous feature changes. Noise was added to the test faces in order to obscure the blurring effect of the morph technique. To determine if the task involves the specific mechanisms used for face recognition, we measured effect of inversion by presenting half of the trials upright and half inverted. Subjects committed far more errors on inverted trials than upright trials. This indicates that this experiment does engage face recognition mechanisms. The prosopagnosic we tested did not show a significant inversion effect.