In phonetics, vowel reduction is any of various changes in the acoustic quality of vowels as a result of changes in stress, sonority, duration, loudness, articulation, or position in the word (e.g. for the Muscogee language), and which are perceived as "weakening". It most often makes the vowels shorter as well.

Vowels which have undergone vowel reduction may be called reduced or weak. In contrast, an unreduced vowel may be described as full or strong. The prototypical reduced vowel in English is schwa. In Australian English, that is the only reduced vowel, though other dialects have additional ones.

There are several ways to distinguish full and reduced vowels in transcription. Some English dictionaries indicate full vowels by marking them for secondary stress even when they are not stressed, so that e.g. ⟨ˌɪ⟩ is a unstressed full vowel while ⟨ɪ⟩ is a reduced schwi. Or the vowel quality may be portrayed as distinct, with reduced vowels centralized, such as full ⟨ʊ⟩ vs reduced ⟨ᵿ⟩ or ⟨ɵ⟩. Since the IPA only supplies letters for two reduced vowels, open ⟨ɐ⟩ and mid ⟨ə⟩, transcribers of languages such as RP English and Russian that have more than these two vary in their choice between an imprecise use of IPA letters such as ⟨ɨ⟩ and ⟨ɵ⟩, or of para-IPA letters such as ⟨ᵻ⟩ and ⟨ᵿ⟩. The French reduced vowel is also rounded, and for a time was written ⟨ᴔ⟩ (turned ⟨œ⟩), but this was not adopted by the IPA and it is now generally written ⟨ə⟩ or occasionally ⟨ø⟩.

Phonetic reduction most often involves a mid-centralization of the vowel, that is, a reduction in the amount of movement of the tongue in pronouncing the vowel, as with the characteristic change of many unstressed vowels at the ends of English words to something approaching schwa. A well-researched type of reduction is that of the neutralization of acoustic distinctions in unstressed vowels, which occurs in many languages. The most common reduced vowel is schwa.

Whereas full vowels are distinguished by height, backness, and roundness, according to Bolinger (1986), reduced unstressed vowels are largely unconcerned with height or roundness. English /ə/, for example, may range phonetically from mid [ə] to [ɐ] to open [a]; English /ᵻ/ ranges from close [ï], [ɪ̈], [ë], to open-mid [ɛ̈]. The primary distinction is that /ᵻ/ is further front than /ə/, contrasted in the numerous English words ending in unstressed -ia. That is, the jaw, which to a large extent controls vowel height, tends to be relaxed when pronouncing reduced vowels. Similarly, English /ᵿ/ ranges through [ʊ̈] and [ö̜]; although it may be labialized to varying degrees, the lips are relaxed in comparison to /uː/, /oʊ/, or /ɔː/. The primary distinction in words like folio is again one of backness. However, the backness distinction is not as great as that of full vowels; reduced vowels are also centralized, and are sometimes referred to by that term. They may also be called obscure, as there is no one-to-one correspondence between full and reduced vowels.

Sound duration is a common factor in reduction: In fast speech, vowels are reduced due to physical limitations of the articulatory organs, e.g., the tongue cannot move to a prototypical position fast or completely enough to produce a full-quality vowel (compare with clipping). Different languages have different types of vowel reduction, and this is one of the difficulties in language acquisition (see e.g. Non-native pronunciations of English and Anglophone pronunciation of foreign languages). Vowel reduction of second language speakers is a separate study.

Stress-related vowel reduction is a principal factor in the development of Indo-European ablaut, as well as other changes reconstructed by historical linguistics.

Vowel reduction is one of the sources of distinction between a spoken language and its written counterpart. Vernacular and formal speech often have different levels of vowel reduction, and so the term "vowel reduction" is also applied to differences in a language variety with respect to, e.g., the language standard.