This paper seeks to illuminate national accounts GDP growth using satellite-recorded nighttime lights in a measurement error model framework. Using recently developed results in conjunction with reasonable assumptions about the exogeneity of the lights data generating process, we identify and estimate the relationship between nighttime light growth and GDP growth, as well as the nonparametric distribution of errors in both measures. We obtain three key results: (i) the elasticity of nighttime lights to GDP is about 1.3; (ii) national accounts GDP growth measures are less precise for low and middle income countries, and nighttime lights can play a big role in improving such measures; and (iii) our new measure of GDP growth, based on the optimal combination of nighttime lights and national accounts data under our identification assumptions, implies that China and India had considerably lower growth rates than official data suggested between 1993 and 2013. We expect our statistical framework and methodology to have a broad impact on measuring GDP using additional information.