This paper explores the encoding of the semantics of evidentiality and indirectivity in some South Asian languages. In my analysis, evidentiality is related to the complex of overlapping categories involving (i) the source of information about an event or state and (ii) its acquisition by an observer/speaker. In some languages several of these notions are morphologically encoded; in others the categories are relatively "covert" and the expression of evidentiality is distributed (Aikhenvald's "scattered") throughout the grammar. The paper summarizes previously published data on inferential systems in Tajik Persian, Kalasha, Khowar, and Nepali, and presents new data on several other languages that have morphologically encoded inferentiality--Yasin Burushaski, three Nuristani languages, and Wakhi. Additionally, other inferentiality-marking strategies are discussed for a cluster of languages including Torwali, Pashto, Shina, and Kohistani, for Hindi and Urdu, and for a cluster of South Indian languages. Evidentiality is known to be highly susceptible to language contact effects Aikhenvald (2003:21-2) and Johanson (2000:81-2). The investigations reported in this paper confirm that evidentialitymarking patterns fall into recognizable areal units and sub-units in South Asia as well. Evidentiality-encoding strategies are seen to group areally with clearly identifiable northern and southern clusters and a mixed area.