This paper reexamines the significant autocorrelation results of foreign currency futures reported by Liu and He [12] in this journal. It argues that extremely thin trading early in the life of individual futures contracts induces unreliable results in [12]. Moreover, the Monte Carlo results clarify the power performance between Lo and MacKinlay's [13] variance ratio tests and Diebold's [3] Q‐statistics; both tests are used by Liu and He.