The Pareto principle is a basic axiom of welfare economics: Improving the well-being of one person without reducing the well-being of any other person improves social welfare. Economists have attempted to extend the Pareto principle to comparisons where population is variable. However, some of these attempts have ignored the non-identity problem. In my presentation I will discuss how the Pareto principle might be extended to the variable population case. This has both implications for population ethics and consequences for optimal population policy in economic growth models.