The current study investigates the intelligibility of face-masked speech while manipulating speaking style, presence of visual information about the speaker, and level of background noise. Speakers produced sentences while in both face-masked and non-face-masked conditions in clear and casual speaking styles. Two online experiments presented the sentences to listeners in multi-talker babble at different signal-to-noise ratios: −6 dB SNR and −3 dB SNR. Listeners completed a word identification task accompanied by either no visual information or visual information indicating whether the speaker was wearing a face mask or not (congruent with the actual face-masking condition). Across both studies, intelligibility is higher for clear speech. Intelligibility is also higher for face-masked speech, suggesting that speakers adapt their productions to be more intelligible in the presence of a physical barrier, namely a face mask. In addition, intelligibility is boosted when listeners are given visual cues that the speaker is wearing a face mask, but only at higher noise levels. We discuss these findings in terms of theories of speech production and perception.