According to Wikipedia, there's a phonemic contrast between /ɕ/ and /ʃ/ in Ubykh, North Qiang, South Qiang and Luxembourgish (though they are merging). Do any other languages exhibit this distinction?
Edit: On a second reading, Wikipedia does not claim that Toda has /ɕ/; I must have misread the article.
You should take any such claims with a large grain of salt. While there is no question that Toda has a contrast, it is not evident that the phonetic realization is [ɕ] vs. [ʃ]. Ladefoged & Maddieson in The sounds of the world's languages 156ff list the Toda sibilants as [s̻ s̱ ʃ ʂ] and [ɕ] is not listed. They also list the Ubykh sibilants as [s ŝ ɕ s̥], without [ʃ] (they state that [ʃ] does not occur in Ubykh and Abkhaz). Although the wiki entry on ɕ lists Norwegian as having that sound, Kristofferson in The phonology of Norwegian lists it as [ʂ], and the wiki entry on Norwegian likewise does not include [ɕ]. Ladefoged & Maddieson put their charcoal where their informants' mouths are and substantiate the claim that Toda has [s̻ s̱ ʃ ʂ]. I argue that to make that specific a claim (that a language has [ɕ] rather than [ʃ] or [ʂ]), one must use some articulatory-measurement method, or, one must be Ian Catford or Peter Ladefoged (both of whom, alas, have another thing in common).
Edit: corrected ɕ/ʂ typo
As a native Polish speaker I would say that @Klaas Edema statement about Polish language was correct.
Polish speakers consecutively tend to classify [ʃ] in English words as [ʂ]. Moreover, in my opinion, in some words newly introduced to Polish were [ʂ] is followed by [i] (in native Polish words "sz" = [ʂ] is never followed by "i" = [i] but by "y" = [ɨ], and "ś" = [ɕ] is never followed by "y" but by "i") like "czipsy" = /ʈ͡ʂipsɨ/ = [t͡ʃipsɨ] (meaning "chips") or when pronouncing some English names with /i/ after /ʃ/ (e.g. "Sheeran") most speakers will actually pronounce it as [ʃ] not [ʂ] (personally I find [ʂi] almost unpronounceable).
Therefore I would say that in Polish [ʃ] is an allophone of [ʂ] while [ɕ] is a distinctive sound. So, given that, it could be said that Polish contrasts /ɕ/ and /ʃ/.
Sanskrit contrasted three sibilants, two of which most probably were [ɕ] and [ʂ] (or [ʃ]). The former was an outcome of an Proto-Indo-Iranian affricate that developed from PIE voiceless palatalised velar stop whilst the latter had its roots in PIE /*s/.
Again here comes the matter of the exact phonetical nature of the second, non-palatal sound, retroflex [ʂ] or post-alveolar [ʃ]. In my view that doesn't matter so much, since Sanskrit didn't contrast all four places of articulation within sibilants (that is dental, post-alveolar, palatal and retroflex). The situation in Sanskrit would be similar to the one in Polish, described by Arsen before. Both [ʂ] and [ʃ] would be allophones of a phoneme that doesn't have to be specifically retroflex or specifically post-alveolar, it just has to contrast with the other two sibilants. In the light of that, it is safe to say that indeed ɕ contrast with ʃ.
The usage of [ʃ] is not consistent according the Wikipedia, with different scholars use it for different sounds because they cannot tell difference between soft and hard consonants.
