This paper presents the main issues involved in estimating potential output. The objective is to describe the alternative methods and analyze their application and implications for growth forecasts and macroeconomic policy in Brazil. The text emphasizes the determinants of potential output under fixed and flexible coefficients of production. Given the wide use of aggregate measures of total factor productivity in growth accounting, and the sensitivity of such a variable to economic assumptions and errors of measurement, the text also presents the main applied critiques and alternatives to aggregate growth-accounting exercises. The main conclusions are: (1) the annual potential growth rate of Brazil’s GDP varies substantially depending on the method and hypotheses adopted and, what is most important, potential GDP is not separable from effective GDP in the long-run; (2) growth-accounting and time-series studies of Brazil result in low potential-output growth rates because they extrapolate the slow growth of 1981-2003 to the future; (3) capital seems to be the main constraint on growth in Brazil and, therefore, a demand-led increase in investment can raise both its effective and potential output levels; (4) however, because of the slow adjustment of the capital stock, an investment boom can also hit a supply constraint before the stock of capital has time to adjust to the growth rate of investment; and (5) aggregate measures of potential output do not carry much information about the economy and, therefore, they should be complemented by sectoral estimates of capacity utilization to identify the bottlenecks in inter-industry flows and the corresponding demand pressures on inflation.