This study explored factors that affect the risk of excluding the support due to recidivism of the recipients within the support period of the housing support projects among the rehabilitation protection projects of the Korea Rehabilitation Agency. The subjects of this analysis were released prisoners who applied and supported by the project from 2005 to September 2009. Continuous survival analysis was used to supplement the limitations of existing logistic regression analysis for the loss of information about the duration of an event (recidivism). The summary of the analysis results is as follows. First, 2.76 % of all recipients of the housing support projects were excluded because of recidivism within the support period. On a three-month basis of analysis, the risk ratio was the highest at 0.0046 (0.46 out of 100) over 7-9 months period. In addition, the cumulative risk ratio of the entire period showed that if the recidivism is not committed for over six years after the start of the support, the likelihood of excluding the support in the future is quite low. Next, the results of the cox regression analysis affecting the risk of exclusion of support due to recidivism showed that health conditions, number of convictions, and the type of committed crime had statistically significant effects. Specifically, those who are in good health, have two or more criminal records, or have a history of village affairs, were found to have increased risk of excluding the support compared to those who are not. Finally, the interpretation of the above results and policy implications were discussed in conclusions and suggestions.