This article aims to account for the phonological phenomenon of the so-called moraic onsets. In the standard theory of moraic phonology as proposed by Hyman (1985), Hayes (1989), Gordon (1999), and Moren (1999), onset segments cannot be associated with moras. This is based on the fact that in many languages, prosodic processes never apply to the onset segments. However, cross-linguistically it is evident that there are many languages and dialects have prosodic processes operate on the onset segments as in Bella Cola, Samothraki Greek, Trukese, Marshallese, Trique, Bellonese, Piraha, Karo, Arabela, Woleaian, Swiss German, Kabardian, Berber, Patani, Kelantan, and Terengganu. The present study attempts to examine the phonology of moraic onset in the Kelantan dialect by using the framework of optimality theory, and the analysis is employing both the prosodic and segmental phonologies. It is apparent that a prosodic word in the Kelantan dialect is equivalent to a foot comprising two syllables under the segmental analysis. This constraint triggers a deletion of initial syllable of an input with three syllables in order to form a disyllabic output. In the standard moraic theory, segmental deletions will leave a floating mora which is then linked to an immediate vowel producing a vowel lengthening. The analysis, however, fails to account for the occurrence of geminate onset as in the word [ssaka] (pusaka) in the Kelantan dialect. In optimality theoretic analysis, the problem can be solved by ranking the constraint NO LONG VOWEL (*LV) higher than NO GEMINATE (*GEM) in the constraint hierarchy. This hierarchical ranking will choose a candidate with germination as an optimal output. The present analysis also suggests that the onset segments can be associated with moras.