Ground measured hourly data from the Solar Radiation Monitoring Laboratory of the University of Oregon over 34 years have enabled a detailed analysis of the relations between direct normal solar irradiance (DNI) and energy yield from concentrating solar thermal power (CSP) generation systems. Six different plant configurations have been selected for simulating the energy yields with System Advisor Model (SAM), based on different parabolic through and central receiver CSP and linear Fresnel commercial plants. The long-term yield energy of each CSP reference plant has been estimated by two different approaches: by modeling a typical DNI year estimated by the Sandia procedure for computing typical meteorological years, or by applying the Sandia methodology directly on the 34 years of hourly gross energy output resulting from multi-year modeling with SAM of the 34 years of meteorological input. Important differences are observed on the typical yield year resulting from each methodology. This work presents thus a complete analysis on how the statistical long-term characteristics of solar resource input might represent the statistical characteristics of long-term yield energy. (C) 2017 Elsevier Ltd. All rights reserved.