We study the problem of preemptive maintenance scheduling for an electricity distribution grid, comprising geographically dispersed assets. We develop scheduling policies that maximize total risk improvement, subject to budget and operational constraints. A main input into developing a schedule is the asset health, which could be represented by time-dependent risk estimates. Constraints include geographical proximity of assets, maintenance and servicing costs (which are assumed to increase as asset health deteriorates), downtime duration, resource constraints and maintenance outsourcing costs. Additionally, the impact of unsatisfied demand, due to an offline asset, is considered.