We tested the capability of deaf adolescents, including a group of users of cochlear implants, to generate inferences during spoken language comprehension and whether they benefited from the use of sign-supported speech (SSS). Stimuli consisted of 24 short video-recorded texts in spoken language and in SSS. Participants responded to literal and inferential multiple-choice questions. In spoken language cochlear implant users had more difficulty in processing inferential than literal information and found predictive inferences harder than associative inferences. The level of spoken language proficiency was related to inference generation, especially for predictive inferences. Similarly, deaf native signers had more difficulties in generating predictive inferences, although SSS increased their comprehension. Lip-reading skills and working memory were positively related to accuracy in SSS. The inclusion of SSS only had a positive impact on the comprehension of native signers. Results suggest that cochlear implant users would benefit from an intervention to enhance verbal skills.