This paper provides a new frequency estimator based on the multiple lags of autocorrelations with discrete Fourier transform (DFT) phase unwrapping. It is proved that this estimator is efficient for short data length while maintaining a low signal noise ratio (SNR) threshold. It is also shown that this estimator is statistically similar to maximum likelihood estimator (MLE), but with lower computational load. Furthermore, the estimator can also be applied to frequency estimation in the presence of moving average (MA) colored noise. In this case, it is shown that the variance of the estimator achieves the Cramer-Rao bound (CRB) asymptotically.