The main purposes of this study are mainly four. They are firstly to explore the correlations between mothers´ childhood attachment experiences, child-rearing behavior and their preschoolers´ attachment quality; secondly to study the differences of mothers´ attachment experiences and child-rearing behavior between high security attached and low security attached children; thirdly to explore whether children´s security score is influenced by mothers´ childhood attachment experiences and their existing child-rearing pattern analyzed by LISREL Model, and finally to explore the suitability of Waters´s method of measurement of attachment of the preschool children named "attachment Q-set". The subjects of this study are 80 three to five year- old children and their mothers. Instruments used in this study are: "Mother-father-peer scale" (Epstein, 1983) which is used to assess mothers´ childhood attachment experiences. "Dyad relationship questionnaire" and "Mothers´ child-rearing behavior scale" are used to assess mothers´ child-rearing pattern. And "Attachment Q-set" (Waters, 1989) is used to assess children´s attachment security score. Results of this study are as follows: 1. Mothers´ childhood attachment experiences, including maternal "independence-encouragement vs overprotection", "acceptance vs rejection" and "maternal idealization" and paternal "independence- encouragement vs overprotection", "acceptance vs rejection" and "paternal idealization", are significantly correlated with each other. 2. Mothers´ child-rearing pattern, including subscale "control", "sensitivity", "acceptance", "guidance", "accessibility´ from "Dyad relationship questionnaire" and subscale "independence-encouragement vs overprotection" and "acceptance vs rejection" from "mothers´ childrearing behavior scale", are significantly correlated with each other except a negative correlation is found between subscale "control" and subscale´ guidance. 3. Almost all variables of mothers´ attachment experiences are significantly correlated with the variables of their child-rearing pattern except a negative correction with subscale "control". 4. Almost all variables of mothers´ attachment experiences are significantly correlated with their child-rearing pattern and their children´s security scores. 5. In order to exploraing significant difference existed between mothers´ attachment experiences, child-rearing pattern of different security attached children, 80 children are diveded into "high secur attached" and "low secur attached" groups according to the sample mean of their security score got from "Attachment Q-set" above or below 1 standard deviation. There were 12 children classified "high securety attached" and 14 children classified "low securety attached". Multiple variable analysis has shown that on the whole, there is a significant difference existed in the mothers´ childhood memory with their own mother and father. T-test result has also shown specific variables within mothers´ attachment experiences have significant differences between high or low secure groups. They are maternal "acceptance vs rejection" and "maternal idealization", paternal "independence- encouragement vs overprotection", "acceptance vs rejection" and "paternal idealization", respectively. 6. No significant difference is existed in the mothers´ childrearing pattern between the high and low secure attached children. 7. Data obtained in this study are further analyzed by LISREL (linear structural relationships) for understanding the relationships among mothers´ childhood attachment experiences, their child-rearing pattern and children´s security of attachment. The result has shown that mothers´ attachment experiences has direct influence on their children´s security of attachment, need not through mothers´ currsect child-rearing paternal. Besides, the direct influence, on the whole effect, parental attachment experiences is more important than maternal of attachment experiences. But on the influence of child-rearing pattern, it seems that mothers´ attachement experiences toward her own mother has more important influence than that toward her own father. According to the results, mothers´ child-rearing pattern the seems not to be a mediate variable between mother´s attachment experiences and their children´s security of attachment. Based on these results, some implications about "Attachment Q-set" are discussed and suggestions or future research are also proposed