Federal policy tools that could affect the concentration of poverty in metropolitan areas have been much advocated but seldom implemented, and their impacts have never been properly measured. This paper estimates the impact of locational constraint and intensity of counseling services on participation by low-income families in the housing voucher program, using data from the Moving to Opportunity experiment. I find that constraining the voucher to low-poverty neighborhoods reduces lease-up by at least 14 percentage points. More intense counseling services raise the lease-up rate, but not enough to overcome the effect of constraint.