In this paper, we present evidence in favour of a syntactic approach to subject drop in Swiss French text messages. Subject drop in our corpus follows patterns found in various so-called "written abbreviated registers" such as diaries, notes etc.: it occurs at the beginning of main sentences and after preposed adjuncts. Based on a corpus of 1100 text messages, collected in 2009/10 (www.sms4science.ch), we test predictions put forward by two approaches to argument drop in abbreviated registers, i.e. the "Avoid Weak Start" hypothesis by Weir (2012a) and the "Truncated CP hypothesis" by Haegeman (2013). While for our data the first approach cannot be excluded, our results more strongly support the syntactic one, despite the fact that some data, especially preposed strong subject moi without clitic resumption, challenge existing analyses. These data suggest that dropped referential subjects can be analysed as instances of familiar topic drop.