There is no positive evidence that Hittite scribes employed the signs including voiced and voiceless stops in a contrastive manner to distinguish between voiced and voiceless values. On the other hand, they probably made consistent use of orthographic single and double stops in intervocalic position to indicate lax and tense qualities or short and long qualities, respectively. From a historical viewpoint single stops continue Proto-Anatolian voiced stops and double stops Proto-Anatolian voiceless stops. The contrast between single and double spellings is generally observed in the case of laryngeals, fricatives and sonorants as well. An immediate question to be raised at this point is whether the same contrast is found between intervocalic single -z- and double -zz-. I will discuss this problem by analyzing the distribution of 3sg. present active endings of mi-conjugation, -Vzzi and -Vzi.As for a-stem and u-stem verbs, the sequence -Vzi is attested side by side with the sequence -Vzzi; e. g., ar-nu-uz-zi “brings” vs. ar-nu-zi. The sequence -Vzi is most naturally explained by the so-called “simplified spellings”. Because the signs az and uz are relatively complicated, Hittite scribes must have spelled the ending without az or uz sign. In general the simplified spellings -azi and -uzi are rare in Old Hittite, but outnumber the unomitted spellings in later Hittite. In any event the alternation between z and zz in the sequences of -a(z)zi and -u(z)zi is of no linguistic contrast.In the case of the -IZ-zi sequence, however, scribes had no motivation for omitting IZ because IZ is a very simple three-stroke sign. Nevertheless, there are a small number of examples in -izi or -ezi with z in Old Hittite texts represented by u-e-mi-zi “finds” KBo VI 2 IV 12 and i-e-zi “does” KBo VI 2 I 60 (both in Old Hittite originals of Law texts). It is noteworthy that these examples all go back to Proto-Anatolian forms characterized by an ending with voiced *d; e. g., u-e-mi-zi<*au-em-ye-di and i-e-zi<*yeh1-di. An obvious inference to be drawn from this fact is that both *-ti and *-di underwent assibilation at a pre-Hittite stage; the ending *-Izzi with zz reflects *-ti and -i/ezi with z reflects *-di, just as geminated stops reflect voiceless stops and single stops reflect voiced stops. I would like to argue that these verbs, though limited in number, preserve a notably archaic feature, resisting to generalization of -Izzi with zz.The above result has a further consequence if we are right in assuming that Proto-Anatolian voiced stops as well as voiceless stops got assibilated before *i. Unlike *t and *d, Proto-Indo-European voiced aspirate *dh never got assibilated in Hittite. The non-assibilation of *dh before *i is supported by the 2sg. imperative i-IT(<*h1i-dhi) “You go!”. Consequently, Proto-Indo-European *d and *dh show different outcomes before *i in Hittite. This will be a piece of evidence for a view that Proto-Indo-European *d and *dh had not been merged at the stage of Proto-Anatolian.