Abstract: In this study, two questions regarding deaf adolescents with skilled and less skilled processing of bridging inference are investigated. 12 deaf readers with skilled and 10 deaf readers with less skilled read short, two-sentence passages that was either a causal sequence or a control temporal sequence, and respond to two questions, one is posited to refer to the relevant knowledge used to evaluate the mediating fact and the other is about explicit information of the discourse. We measured correct responses and answer times to questions, as well as recorded the eye-movements of the participants while reading. Analysis of the skilled participants’ answer times to questions were consistently faster in the causal condition than the control condition, indicating skilled adolescents with deaf construct bridging inference online. We can’t find the validation effect in less skilled adolescents with deaf, indicating they don’t construct bridging inference online. However, the eye-movements data revealed that the less skilled readers with deaf spent more time fixating, made more fixations, and made total times in the causal sentences while reading than did skilled readers with deaf . One explanation is that they make the inferences, but are unable to retain them offline or they make the inferences but are unable to incorporate them into meaning formed from text and prior knowledge activated during the reading process. The study also suggests that skilled deaf readers draw bridging inference automatically and efficiently, less skilled deaf reader are less efficient and are not up to the degree of automation at drawing bridging inference. Implication of these finding and suggestions for future research are offered.