Internal working models are based on young children’s expectations for the behavior of their attachment figures that develop into broader representations of their relational experiences, and decision rules about how to interact with others. These internal working models are thought to influence the child’s subsequent processing of social experiences and allow the child to anticipate, plan for, and adapt to his or her social world. Current research studies largely concentrated on the internal working models of attachment in childhood and adulthood; very little inquiry has been conducted on internal working models of preverbal infants and toddlers. It was not until 2007 that a study by Johnson, Dweck and Chen had found evidence for infants’ internal working models of attachment using a habituation paradigm. In addition, some theorists and researchers have indicated that maternal sensitivity is associated with internal working models in childhood; and also it has been suggested that maternal sensitivity may affect toddler’s attachment status. Therefore, internal working models may play an important role in relationship between maternal sensitivity and toddler’s attachment status. However, there is inadequate research on maternal sensitivity and internal working models in infancy. The present study explored the relationship between toddlers’ attachment status, maternal sensitivity and internal working models of toddlers. We hypothesized that toddlers of different attachment status might have unique patterns of expectations about their mother’s responsive behavior, and the expectations might correspond to maternal sensitivity in caregiver-infant interactions. Forty-one toddlers (M = 14.01 months, SD = 0.81) and their mothers from middle class backgrounds participated in the study. A visual habituation study using abstract animations with separation and reunion events was conducted to test a part of the toddlers’ internal working models—their expectations of a caregiver’s responsiveness. Toddlers’ attachment status was assessed by the Strange Situation procedure. Maternal sensitivity in caregiver–toddler interactions was measured by Maternal Behavior Q-Sort. The data was analyzed using repeated-measure analysis of variance, chi-square analyses, paired t tests, q-factor analysis and LCA. Results were as follows: (1) Toddlers’ expectations of a caregiver’s responsiveness depended on their own attachment status: insecure toddlers expected their caregivers to withhold comfort; (2) Toddlers’ expectations of a caregiver’s responsiveness depended on their mother’s behavior: the toddlers of Self-centered mothers expected caregivers to be unresponsive; (3) The results found by LCA indicated that there were two types of attachment system: Category 1, the toddlers of Child-centered mothers tended to expect their caregivers to be responsive, and showed secure attachment behaviors. Category 2, the toddlers of Self-centered mothers tended to expect their caregivers to be unresponsive, and showed insecure attachment behaviors in Strange Situation procedure. These results indicated that it was effective to using a visual habituation procedure to measure internal working models of preverbal toddlers. The claims that toddlers’ internal working models play an important role in the relationship between maternal sensitivity and toddler’s attachment status can be supported. If we want to change the attachment status of toddlers through their internal working models, we should pay attention to the behavior of their mothers.