This paper offers a synchronic rule set to fully account for the standard sandhi at word boundaries in classical Sanskrit. Although Sanskrit is often characterized as having a phrasal phonology that is unnatural to the point of artificiality, this rule set shows that phonologically unnatural rules can be isolated in the lexical phonology, where one expects to find rules that require historical explanation. It also shows that with the exception of three rules that have lexical conditions, the rules of external sandhi can be described as syllable-juncture P2 rules rather than as word-juncture rules. Thus they obey Kaisse’s theory of sandhi despite her original analysis. It further shows that pada suffixes, which have traditionally been held to exceptionally obey external sandhi, are not exceptional at all when syllable structure is taken into consideration. A computer program has been developed for testing the rule set over a test suite, and a printout showing the derivation of representative word pairs is included.