In this paper, we develop the multipower estimators for the integrated volatility in (Barndorff-Nielsen and Shephard in J. Financ. Econom. 2:1-37, 2004); these estimators allow the presence of jumps in the underlying driving process and the simultaneous presence of microstructure noise and multiple records of observations. By multiple records we mean more than one observation recorded on a single time stamp, as often seen in stock markets, in particular, for heavily traded securities, for a data set with even millisecond frequency. We establish the consistency and asymptotic normality of the estimators for both noise-free and noise-present cases. Simulation studies confirm our theoretical results. We apply the estimators to a real high-frequency data set.