peakers often produce simple up-and-down flicks of the hands, called beat gestures, that are carefully timed with speech. On the sentential level, beat gestures emphasize words that provide important information. On the word level, these gestures typically fall on stressed syllables. A recent lexical-decision study showed that beat gestures speed up the lexical access of words in sentences. However, this effect was not modulated by the temporal alignment on the word level (henceforth, stress-congruency), as beat gestures falling on stressed and unstressed syllables facilitated lexical access to the same extent. The present study tested i) whether beat gestures’ facilitatory effect on lexical access can be replicated in an online setting and ii) whether a larger facilitation and a stress-congruency effect might emerge when listening to speech-in-noise, a condition in which listeners weigh more heavily the visual cues provided by speakers. While no stress-congruency effect emerged, beat gestures’ facilitatory effect on lexical access was successfully replicated and shown to be larger when listening to speech-in-noise, even improving lexical-decision accuracy, compared to speech-in-quiet. These findings confirm that larger audiovisual effects are observed when speech is degraded, providing further insights into the role of beat gestures in multimodal communication.