The selection of a one or two-tailed t-test must be made before the experiment is performed. It is not “cricket” to find a that tOBS = -1.92, and then say “I really meant to do a one-tailed t-test.” Because reviewers of articles submitted for publication are sometimes suspicious when a one-tailed t-test is done, the recommendation is that if there is any doubt, a two-tailed test should be done.（

來自統計書本）

總結一下：

1，有很強的證據表明一方比另一方大或者小（有大量的先驗知識）；
2，只對一側的資料感興趣。
其它情況都要用雙尾。