About 55 percent of refugees placed in Utica, New York plan to live there for more than 10 years, while roughly one in five refugees plan to leave within 5 years. Unlike economic immigrants, refugees have little say on where they are resettled in the United States. Federal policy does not prevent refugee secondary migration, moving from their initially assigned community, but since resettlement funding flows to the location and does not follow refugees who move, there is a strong policy bias for sedentary assimilation. For example, under current policies, funding for housing and language acquisition flow through selected agencies in the assigned community based on the number of refugees settled in each time period. Despite the clear policy biases, very little is known about either the share of refugees who leave their assigned community or the factors that compel them to move, primarily due to the lack of U.S. data on refugees after they are settled. Using the Survey of Utica Refugee Retention and Financial Inclusion (SURRFI), the authors estimate models of refugee retention and migration for a sample of more than 600 refugee households. The refugees in the sample come from over 30 countries and were resettled in Utica, New York over a 35 year time span. These new data allow the authors to test a variety of migration theories including spatial assimilation, enclave and segmented assimilation, human capital and barrier theories. The authors capitalize on questions of desired duration of stay combined with data on exits of family members. Preliminary probit model estimates show that, other variables held constant, the probability that a refugee moves from the city of initial settlement does not depend on household income but shows strong time patterns with exits increasing with time but at a decreasing rate. Muslim refugees are less likely to move away or have family members move away. Similarly, while Muslim households are more likely to report a desire to stay long term, refugees with no religious affiliation are less likely to stay. Finally we find that generation 1.5 refugees, defined as those entering under the age of 10, are far less likely to want to stay an additional 10 years than those who came when older than 10 years of age. We are currently refining and estimating new models.