The idea that conscious awareness of a recognizable image is necessary for it to evoke the recognition potential (RP) was tested by asking bilingual subjects to selectively attend to superimposed English and Chinese word images. The subjects detected most of the words in the attended language, but were largely oblivious of words in the non-attended language. Attended word images evoked the RP. Non-attended words did not. RP latency was less for Chinese than for English words. This provided a basis for inferring which language a subject was trying to read when valid English and Chinese words were both present. A subject was looking for Chinese if the latency was short and for English if it was long. The results showed that selective attention had a powerful effect on the RP. They supported the idea that conscious awareness is necessary for evoking it, though they did not rule out the theoretical possibility that some method not yet tested could be found that would block conscious awareness without blocking the RP. The sensitivity of the RP to what a subject is trying to see and its low variance seem to provide advantages for studying visual perception. It provides a short latency indicator of image processing that merits further investigation. Use of it may lead to a better understanding of visual perceptual processes.