*Toshimasa Yamazaki, *Yoko Takaki, and *Ken‐ichi Kamijo *NEC Fundamental Research Laboratories, Tsukuba, Japan . Purpose: We have developed an adaptive temporal filtering system (ATFS) to prevent health hazards due to flicker stimuli with high luminance from televisions (TVs), videos, and other displays. The characteristics of this ATFS are to calculate risk index from images based on the human visual function model as temporal frequency tuning, to transform high‐risk images into those with low‐risk indices in real‐time by temporal low‐pass filtering, and to output normal images as they are. The risk index is defined by the difference in temporal frequency amplitude between the model response to input images and the original images. According to the risk index, images are adaptively and temporarily blurred. This system could reduce frame‐to‐frame or field‐to‐field flicker stimuli at ∼10–30 Hz, frequencies that are particularly risky for photosensitive individuals. When the ATFS receives image signals from videos, tuners, or games, normal images are presented as such, and the high‐risk ones, as low‐pass–filtered images on the TV monitor. For 13 patients, a filtered scene by an ATFS with low efficacy elicited a generalized photoparoxysmal response (PPR) in six patients, whereas that by an ATFS with high efficacy elicited no PPRs (Epilepsia 2002;43:530–4). To investigate the effects of these two kinds of the ATFS on normal subjects, we conducted a subjective assessment experiment. Methods: Subjects were 26 volunteers (nine female and 17 male subjects; mean age; 36.96 years). The ATF effect depends on the slope of the temporal response–modulation function of the model. Two kinds of slopes were used in the present experiment. “Strong” and “mild” effects correspond to steep and gentle slopes, respectively. Fifteen kinds of images, each of which has three different ATF effects including the original, were presented. Examples of images with high‐risk index were those of a person exposed to a flash light of cameras, a high‐luminance photograph, and a geometric pattern being continuously switched. The images were presented on a 9‐inch color monitor that was placed at a distance of 60 cm from the subjects. The angle between the horizontal line and the subjects’ eyes was 10 degrees. The illumination of the room was ∼ 1,000 Lux. The subjects scored subjective feeling for the images. The evaluation items consisted of subjective feelings such as glaring, flickering, unnatural and blurred, and physical symptoms such as “the eyes hurt” and “want to turn the eyes away.” Results: The “strong” ATFS reduced “glaring”[t(28) = 1.59, p = 0.062] and “flickering”[t(28) = 1.64, p = 0.056] scores in comparison with the originals. Conversely, the “strong” ATFS increased “unnatural”[t(28) = 2.23, p = 0.017] and “blurred”[t(28) = 3.65, p = 0.001] scores, compared with the “mild” ATFS. The physical symptoms were not greatly affected by the ATF effect. Discussion: For normal subjects also, our ATFS proved to be effective to neutralize images with violent luminance changes, as supported by the present subjective assessment of the ATFS. Future work should address the content dependency of the ATF effects and specify the appropriate ATFS parameter set for individuals, which determines the slope. Acknowledgment: This study was supported by Millennium Project of the Ministry of Education, Culture, Sports, Science and Technology (2000–2002).