To explore the binaural integration abilities of six educable mentally retarded boys (ages 8 to 13) and six normal boys (ages 7 to 12) to detect possible brain inju"y, an adaptation of Matzker's (1958) technique involving separating words into high and low frequencies was used. One frequency filter system presented frequencies from 425 to 1275 cycles per second (Hz) in one band and 2550 to 6800 Hz in the other; the second system presented frequencies from 637 to 1275 Hz and from 2550 to 5100 Hz. Subjects were asked to repeat 10 words from Thorndike's 1929 word list presented in each of four conditions: high frequencies only, low frequencies only, high and low frequencies to separate ears (integration), and both frequencies to both ears (normal). The integration and binaural normal conditions were the only two conditions useful for assessing auditory integration. The second system produced higher mean errors under each condition, but no significant differences were found between the retarded and normal groups in mean number of errors. Only one retardate and possibly one normal boy were suspected of even minimal brain injury. Suggestions are made for future research. (RP)