I investigate how verbal aspect was expressed in multiple Sanskrit and Middle Indic language varieties andwhat changes in aspectual semantics accompanied the radical changes in verbal morphology that occurred
over the many centuries separating Vedic Sanskrit from late Middle Indic. My discussion is based on data
extracted from text samples representing each variety covered. I collected from each sample all verbal
predicates with past reference (or encoding states caused by past events) and identified the viewpoint aspect
of each form wherever possible. This enabled me to describe the morphological expression of aspect in
each linguistic stratum under consideration. Furthermore, I discuss and suggest theoretically informed
explanations for the diachronic changes observed. The major predicate categories observed in Sanskrit are
the imperfect, perfect, aorist, past participle and past-referring present. The imperfect and perfect (earlier
distinct) were both aspectually neutral at the Vedic prose stage and so remained. The aorist encoded
anterior aspect in Vedic prose, becoming compatible with perfective and imperfective readings in post-Vedic
Sanskrit (so merging semantically with the imperfect and perfect). In Middle Indic the merger was both
semantic and morphological, since only a single aspectually neutral finite past, itself lost in later varieties, is
attested. I suggest that the aspectual merger of the finite past tenses in post-Vedic Sanskrit may be due to
Middle Indic influence. The past participle, like the aorist, was an anterior category in Vedic prose and later
underwent semantic generalisation, since it has perfective readings in all later Sanskrit and Middle Indic
corpora; however, generalisation never proceeded to a point where the participle became compatible with
imperfective readings. I argue that the past-referring present, originally aspectually neutral and still neutral
in Vedic prose and at least one early Middle Indic variety, later became associated with imperfective aspect
(in both Sanskrit and Middle Indic) due to blocking of non-imperfective readings by the past participle. I
also draw attention to the presence of perfective-looking presents even in later forms of Middle Indic and
discuss semantic and contextual regularities among them, especially a high frequency of verbs denoting
speech events.