We present results of a comparative acoustic analysis of pre-pausal glottalisation in Ṣan‘āni, the Arabic dialect of the old city of Ṣan‘ā, and Mahriyōt, an eastern Yemeni dialect of the Modern South Arabian language Mehri.¹ Data are analysed from one speaker of each variety. In the obstruent series, both varieties maintain the three-way voiced – emphatic – voiceless contrast of Proto-Semitic. In Ṣan‘āni, sonorants and voiced and emphatic obstruents glottalise pre-pausally, while voiceless aspirated stops pre-aspirate, leading to neutralisation of the laryngeal contrast between voiced and emphatic obstruents. Our analyses of Ṣan‘āni demonstrate that while oral stops and vowels post-glottalise, other segments tend to pre-glottalise and are prone to lenition, particularly the (non-sibilant) coronals. In Mahriyōt, emphatic and voiced obstruents are glottalised pre-pausally, and voiceless aspirates are heavily post-aspirated. Sonorants and fricatives may be pre-glottalised, but, in contrast to Ṣan‘āni, no lenition is evident. Results also show that while the Mahriyōt velar emphatic is ejective in all positions, the other emphatics become ejective only pre-pausally.