In this paper, we analyze the empirical likelihood under strongly stationary negatively associated (NA) series by using the blockwise technique. Our results show that the statistics is asymptotically chi-square distributed and that the corresponding confidence interval can be constructed.