People usually show a stable preference for one of their eyes when monocular viewing is required (‘sighting dominance’) or under dichoptic stimulation conditions (‘sensory eye-dominance’). Current procedures to assess this ‘eye dominance’ are prone to error. Here we present a new method that provides a continuous measure of eye dominance and overcomes limitations of previous procedures. We presented dichoptic streams of randomly selected alphanumeric characters at rates around 5 Hz and asked observers to detect a particular character. In most subjects, the dichoptic streams of letters did not perceptually overlap, instead many participants were never aware that two letters were always presented. Interocular differences in target detection were evident in most observers, thus targets presented to one eye were always detected while targets presented to the other eye were generally missed. These interocular differences (i.e., eye dominance), were normally distributed and showed high test-retest reliability.