It is important to know how aggregate economic growth or contraction was distributed according to initial levels of living. In particular, to what extent can it be said that growth was "pro-poor?" There are problems with past methods of addressing this question, notably that the measures used are inconsistent with the properties that are considered desirable for a measure of the level of poverty. The authors provide some new tools for assessing to what extent the aggregate growth process in an economy is pro-poor. The key measurement tools is the "growth incidence curve," which gives growth rates by quantiles (such as percentiles) ranked by income. Taking the area under this curve up to the headcount index of poverty gives a measure of the rate of pro-poor growth consistent with the Watts index for the level of poverty. The authors give examples using survey data for China during the 1990s. Over 1990-99, the ordinary growth rate of household income per capita in China was 7 percent a year. The growth rate by quantile varied from 3 percent for the poorest percentile to 11 percent for the richest, while the rate of pro-poor growth was around 4 percent. The pattern was reversed for a few years in the mid-1990s, when the rate of pro-poor growth rose to 10 percent a year--above the ordinary growth rate of 8 percent.