Small-world structures of human brain functional networks can promote the information forwarding and integration among brain regions with low energy consumption and connection cost. However, traditional metrics have the problem of low accuracy during‘small-world-ness’measurement of human brain functional networks. In order to solve this problem, we proposed a‘small-world-ness’metric named efficiency based small world(ESW) index with the consideration of the high efficiency characteristic of human brain. The validity of the ESW on simulation networks under different edge densities and node scales is verified by comparing with the traditional measurement including clustering coefficient, characteristic path length, new small world index sigma and omega. Simultaneously, the‘small-world-ness’of real world human brain functional networks is verified. The variation on the‘small-world-ness’of healthy volunteers(HV) with different thresholds is found firstly. Then, by comparing and 1analyzing the‘small-world-ness’of HV, patients with childhood onset schizophrenia(COS) and patients with attention deficit hyperactivity disorder(ADHD), it is showed that the‘small-world-ness’of COS and ADHD patients weakens. The study on the small-world-ness of human brain functional networks provides a reference for the relevant researchers, which means we can explore the working and connection patterns of human brain from the aspect of network topologies.