Sensory integration (SI) is the ability to organize and process information from the different sensory channels and to interrelate and synthesize these inputs in order to emit an adaptive motor response. Young children with SI deficits would exhibit delays in fine and gross motor skills, poor balance, uncoordination, developmental delay and poor skills in physical activities. Commercial tests are available (e.g., DeGangi-Berk Test) to provide measures on the three sensory integration sub-domains (i.e., postural control, bilateral motor integration, and reflex integration) for preschool children. A careful review of the past research indicates that the three sub-domains have been analyzed either independently or together with total scores that are derived by adding the scores of the three domains. The limitation of these analyses was that the compositional pattern of the three sub-domains could not be described and the pattern changes could not be detected. The purpose of this study is twofold: (a) use ternary diagram technique to describe the compositional pattern of the three SI sub-domains, and (b) use partial least squares technique to simultaneously investigate the effect of the "sensory integration protocol" intervention to the SI sub?domains. Two intact preschool classes (n1=15 and n2=16) served as the experimental and control groups with that the experimental group had the sensory integration protocol intervention and the control group had not (i.e., 2 (groups) x 2 (time) repeated measures on the time factor design). The pre- and post-treatment data were collected using the DeGangi-Berk Test of Sensory Integration (De-Gangi, 1979) for both groups. The results showed a significant improvement in the total score of the sensory integration in the experimental group (p<0.01). However, the examination of the ternary diagrams indicated that these children with learning disabilities had higher deficit of reflex integration performance than that of postural control and bilateral motor integration performances. The result of partial least squares analysis showed that the sensory integration protocol intervention had no significant effect (p=0.83) on the deficit pattern. The results clearly indicated that the reflex integration was performed worst among the three sub-domains. These results suggested that further efforts should be made to improve the reflex integration for these children so that their learning ability can be enhanced substantially. The results implicate that the sensory integration sub-domains should be simultaneously assessed so that the weakest component among the three sub-domains could be detected and the specific efforts could be made to substantially improve their learning ability.Keyword(s): adapted physical activity, assessment, early childhood