This paper presents a non-parametric method for testing the hypothesis that asset prices have continuous sample paths using high-frequency data. The authors introduce bipower variation (BPV), a measure that can help distinguish between continuous price movements and jumps. They show that BPV can be used to consistently estimate the continuous and discontinuous components of quadratic variation. The paper also develops asymptotic distribution theory for tests based on BPV and quadratic variation, which are used to test for jumps in financial time series. The authors apply their tests to simulated data and real exchange rate data, finding that jumps are frequently rejected in the null hypothesis of continuous sample paths. They also relate some of the detected jumps to macroeconomic announcements. The paper concludes that BPV is slightly less efficient than quadratic variation as an estimator of quadratic variation in the case of continuous sample paths, but it provides a robust method for testing for jumps in financial data.This paper presents a non-parametric method for testing the hypothesis that asset prices have continuous sample paths using high-frequency data. The authors introduce bipower variation (BPV), a measure that can help distinguish between continuous price movements and jumps. They show that BPV can be used to consistently estimate the continuous and discontinuous components of quadratic variation. The paper also develops asymptotic distribution theory for tests based on BPV and quadratic variation, which are used to test for jumps in financial time series. The authors apply their tests to simulated data and real exchange rate data, finding that jumps are frequently rejected in the null hypothesis of continuous sample paths. They also relate some of the detected jumps to macroeconomic announcements. The paper concludes that BPV is slightly less efficient than quadratic variation as an estimator of quadratic variation in the case of continuous sample paths, but it provides a robust method for testing for jumps in financial data.