Power spectral density (PSD) analysis is an important part of understanding line-edge and linewidth roughness in lithography. But uncertainty in the measured PSD, both random and systematic, complicates interpretation. It is essential to understand and quantify the sources of the measured PSD’s uncertainty and to develop mitigation strategies. Both analytical derivations and simulations of rough features are used to evaluate data window functions for reducing spectral leakage and to understand the impact of data detrending on biases in PSD, autocovariance function (ACF), and height-to-height covariance function measurement. A generalized Welch window was found to be best among the windows tested. Linear detrending for line-edge roughness measurement results in underestimation of the low-frequency PSD and errors in the ACF and height-to-height covariance function. Measuring multiple edges per scanning electron microscope image reduces this detrending bias.