This paper describes and analyzes vowel hiatus resolution in Kikuyu, filling empirical gaps in previous descriptions and addressing differences between our dataand earlier published data that may reflect dialectal and/or generational differ-ences. We demonstrate that Kikuyu’s superficially very complex system of vowelhiatus resolution can be analyzed straightforwardly using ordered autosegmentalrules, most notably ATR shift, which captures the observation that, with limitedsystematic exceptions, when two vowels come together with underlying features[–atr][+atr] or [–atr][–atr] they surface as [+atr][–atr], while underlying sequences [+atr][–atr] and [+atr][+atr] surface unchanged (in other words, if either vowel is [–atr] underlyingly, the surface form will be [+atr][–atr]). Otherrules include fusion, assimilation, diphthongization, and shortening.