This paper develops the idea of renewal time sampling, a novel sampling scheme constructed from stopping times of semimartingales. Based on this new sampling scheme we proposes a class of volatility estimators named renewal based volatility estimators. In this paper we show that: (1) The spot variance of a continuous martingale can be expressed in terms of the conditional intensity or conditional duration density of renewal sampling times; (2) In an infill asymptotics setting, renewal based volatility estimators are consistent and jump-robust estimators of the integrated variance of a general semimartingale; (3) Renewal time sampling and range-based sampling have a higher sampling efficiency than equidistant return-based sampling.