Recently a great interest has been put on a captioned system for the deaf. However, it is still quite unclear under what optimum conditions the deaf show proficient performance. At present, methods of producing captioned television are still in the experimental stage. The purpose of this study was to determine the conditions favorable in producing captioned TV. The subjects consisted of deaf students in elementary grades 4-6, and secondary grades 1-3, and hearing students at the elementary level in the fourth grade. Two experiments were used in the study. In both experiments, procedures were similar. The subjects were tested in groups, seated 3-5 meters away from a television monitor. Afterviewing a captioned program, they were given a comprehension test, consisting of a descriptive section and a multi-choice section; the story was selected from materials for elementary school children viewers by NHK(NihonHoso-Kyokai). In the first experiment, 69 elementary-level deaf-school children, 91 secondary-level deaf-school students and 39 elementary school children participated. Thisexperiment utilized a 2×3×2 factorial design to assess the effects of caption language level(word, sentence), caption presentation time(3 sec, 5 sec, 7 sec), and developmental level(elementary, secondary)on the students. The obtained results were as follows; the developmental level was found to significantly affect comprehension of a story on captioned TV. However, caption presentation time and caption language level were not to be found to have a main effect. Significant interaction occurred between caption language level and developmental level. In the second experiment, the same subjects participated. This experiment utilized a 3×2×2 factorial design to assess the effects of caption language level(word, simple sentence, complicated sentence), captioned rate(30%, 60%), and developmental level(elementary, secondary). The obtained results were as follows; the developmental level was found to have a main effect, however captioned rate and caption language level were found to have no effect. Still, significant interaction occurred between developmental level and captioned rate. These results suggested that the optimum captioned conditions in producing captioned TV for the deaf should be using a word script of 60% of the full text at the elementary level, and sentences of 30% at the secondary level. Finally a direction for further captioned research and its related problems were discussed.