Abstract This study examines the degree of time persistence in U.S. disaggregated renewable energy consumption (hydropower, geothermal, solar, wind, wood, waste, and biofuels) using innovative fractional integration and autoregressive models with monthly data for the period 1994:2 to 2011:10. The results indicate that in the case of hydropower, solar, wind, waste, and biofuels the estimates of fractional integration are higher than 0.5 but less than 1.0 implying nonstationary, but mean reverting behavior. In the case of geothermal and waste the estimates of fractional integration are around 0.5 and in the boundary case between stationarity and nonstationarity. For wood, the estimate of fractional integration is significantly smaller than 0.5 and thus showing stationary behavior with long memory behavior. Furthermore, the study incorporates the presence of breaks in the data with the absence of breaks in hydropower, geothermal, solar, wind, wood, and biofuels, but a single break in the case of waste due to the inclusion of non-renewable waste from non-biogenic sources through 2000. The results reveal that U.S. disaggregated renewable energy consumption measures are better explained in terms of a long memory model that incorporates persistence components and seasonality.