This paper aims to examine the ethnic variations of condolence messages written by Malaysians. Data comprise 36 authentic SMS condolence messages written in English by predominantly 47% Chinese, 33% Malays and 20% Indians. Data were retrieved from the recipient and then projected as a word document. They are then coded according to ethnicity, age and gender and the number of words were counted. The speech act theory is applied to determine the expressive functions and this is then followed by a linguistic analysis. Analysis suggests that the SMS condolences can be classified into 13 expressive functions. The condolences written by Chinese writers expressed more hope and sympathy; those written by Malay writers were more indirect and polite and they expressed more apologies while those written by Indian writers expressed more God-related statements. Since the findings are derived from a small set of data the outcome of this paper cannot be generalised. Nonetheless, they are authentic data which are difficult to access hence, these findings will serve as a window for other cultures to understand the ethnic differences of multicultural Malaysia. Indeed, the outcome of this paper can contribute to a better understanding of intercultural communication.