One of the most controversial issues in Korean phonology seems to be the sai-sios phenomenon in compound words. Cook (1987) states that "hundreds and perhaps thousands of pages could have been spared had sai-siosNOT been written in the first place." Sai-sios has been described in terms of t-epenthesis, boundaries, brackets, gemination, strata in lexical phonology, C-slots in CV phonology, empty X-slots, or a laryngeal feature postulated as a denominal adjective morpheme. Most of the phonological analysis of sai-sios is based on data from Modern Korean. This paper attempts to account primarily for consonants inserted in noun compounds in fifteenth-century Korean. I am not concerned with the syntactic functions of sai-sios. I will examine noun compounding as reflected in two publications of the fifteenth century, the Pŏnyŏk Nogŏltae (1510?) and the Nogŏltae Ŏnhae (1670). Following Baek (1991), I assume that an articulatory pause accompanied by a glottal stop occurs between the two members of noun compounds and that sai-sios has provided a convenient means to represent this articulatory pause. I will show that this articulatory pause gives rise to a range of phonetic variations and processes both in fifteenth-century Korean and in Modern Korean, processes such as t-epenthesis, glottalization, gemination, and others.