The paper describes how micro-simulation techniques were used to forecast relative child poverty in the UK in 2010 and 2020 under various scenarios for future policy and socio-demographic change. This modelling work was at the core of the Joseph Rowntree Foundation's project 'What will it take to end child poverty?'. It starts by looking at what would happen under present policies, projecting that the poverty rate would be fairly stable, and fall substantially short of meeting government targets for reduction. It then looks at the possible structure and cost of various benefit and tax credit changes, whose projected effects would be sufficient to meet the targets. These projections are not forecasts, but illustrations that allow the size of the impact of various policies to be compared, holding constant a number of unknown factors that could influence the eventual poverty rate.