Working memory capacity involves the ability of temporary storage and simultaneous manipulation of information. Baddeley and Hitch proposed that working memory could be divided into three subsystems, phonological loop concerned with storage of verbal and acoustic information, a second subsystem, the visuo-spatial sketchpad providing its visual and spatial equivalent, and the central executive, a domain general system involved attention of focus, divided attention, attention switching and communication with long term memory. Substantial studies suggest that the structure of working memory in children is consistent with Baddeley’s 3-subsystem model. Working memory is believed to be essential for a very wide range of complex cognitive tasks. Complex span tasks, like reading span, listening span, operating span, counting span, are usually used to measure working memory capacity development. However, the complex span tasks have not been fully understood and there has been a heated controversy about what they really measure. The aims of the current study was to examine the development of working memory through testing the function of working memory components based on Baddeley’s model and to explore how these measurements were related to the complex span, commonly acknowledged as index of working memory capacity. 225 pupils aged from 6 to 9 years old performed nine working memory tasks, including digit span, non-word span, visual pattern span, spatial span, focus of attention, divided attention, switching attention, semantic fluency test and listening span task, to be examined the capacity of phonological loop, visuo-spatial sketchpad and central executive respectively. The results showed phonological memory, visuos-patial storage and central executive functions grew with age at changing pace during 6 to 9 years of age. Post hoc tests demonstrated that phonological storage developed slowly from 6 to 8 and increased significantly between 8-9 years of age, while visuo-spatial capacity experienced significant increase from 6 to 7 , 7 to 8 years old and showed no significantly differnces between 8 and 9 years age. The results also suggested that three central executive functions , retrieval of long term memory, divided attention and focus of attention, witnessed significant increase between 6 and 7, 8 and 9 years of age, while switching attention ability increased significantly from 6 to 7 years of age. SEM analysis showed that both phonological loop and central executive positively affected the listening span. The overall pattern of results has important implications for understanding and measurement of working memory development and structure in childhood. we conclude that both the capacity of the 3 subsystems of working memory and the relationship between them are changing during childhood, and both the central executive functions and phonological loop are involved in verbal working memory tasks.