|Abstract:||The goal of this paper is to compare two different hypotheses about the insertion of morphs and allomorphy in Optimality Theory. One of them, the Morphs through Constraints Hypothesis (MCH) claims that the phonological realization of morphemes (morphs) is introduced through language-particular constraints. The other hypothesis, the Morphs in the Input Hypothesis (MIH) claims that the inputs to GEN contain all the relevant phonological information. It is shown that the MIH is clearly superior to the MCH in accounting for voicing neutralization in languages like Catalan. The two hypotheses seem to fare even in dealing with other phenomena, such as phonologically-conditioned allomorphy or OCP-triggered epenthesis vs. haplology in English possessives and plurals. Finally, although the MCH seems to be a simpler hypothesis for lexical exceptions, it is shown that, when certain aspects are taken into account, it runs into problems.