An amendment to the domestic violence prevention act in Taiwan has given the police more authority to interfere domestic violence case more zealously, in order to offer more protection for the victims, and deter offenders from recidivism. However, police response concerning domestic violence has not improved as expected, and therefore an empirical research based on the influential factors will be necessary. Based on coding and analyzing the incident reports of domestic violence and the child protection cases, the current study intended to develop a fuller picture of police dealing with domestic violence cases and explore the relationship between the characteristics of organization and the persons involved and the incident. Some recommendations for the policy and practice were also addressed. According to the results of previous studies and the reports of incidents, the present study focuses on investigating the contextual, demographic, and situational effects to active police responses, namely the petition of emergent protection order and arrest the abuser. The factors are classified into five aspects: organizational factors of police authority, the characteristic of offender, victim, incident and the attitude of victim. The present study employs data from 2,202 processed domestic violence cases, including 276 cases with protection order applied by the police, and 33 cases with offenders arrested. Data were collected from 17 different branches in two Taiwanese counties. To sieve out the significant influential factors, I used chi-square test to test the independence. In the end, two logistic regression models were developed to explain the significant variables of the police response regarding the domestic violence cases. The finding shows that the significant variables of police responses in enforcing criminal protection order and those of arresting the suspect are largely dissimilar. However, the two responses are influenced by the domestic violence history, drug usage, physical injury, and victim’s willing of proceeding with cases. It is also found that the victim’s and case’s characteristics are not significant in the final logistic regression model. The regression model also shows that the victim’s proceeding with cases is the most significant variable. Though the police response will take accounts of the domestic violence prevention act at large, the victim’s attitude still play the key role. The police seldom use their professional discretion to estimate the dangerousness of the abuser and the risk of re-victimization of the victim. Some suggestions are drawn from the research findings. Firstly, Police Prevention Domestic Violence Guide Book should include clear decision making details for criminal orders of protection and arrest. Secondly, the police should receive a standard and consistent training and instruction before dealing with DV cases to minimize diversity. Thirdly, the domestic incident report should include the police decision making record and the offender’s interview record. Finally, it is suggested that more efforts are needed to integrate intelligence system of the police, other criminal justice agency, and third parties. To achieve the aim of domestic violence prevention act, an integrative system to support the victim and prevent them from further harm is desperately needed.