The current study sought to examine the utility of intra-individual variability (IIV) in distinguishing participants with prodromal Huntington disease (HD) from nongene-expanded controls. IIV across 15 neuropsychological tasks and within-task IIV using a self-paced timing task were compared as a single measure of processing speed (Symbol Digit Modalities Test [SDMT]) in 693 gene-expanded and 191 nongene-expanded participants from the PREDICT-HD study. After adjusting for depressive symptoms and motor functioning, individuals estimated to be closest to HD diagnosis displayed higher levels of across- and within-task variability when compared to controls and those prodromal HD participants far from disease onset (F ICV(3,877)=11.25; p<.0001; F PacedTiming(3,877)=22.89; p<.0001). When prodromal HD participants closest to HD diagnosis were compared to controls, Cohen's d effect sizes were larger in magnitude for the within-task variability measure, paced timing (-1.01), and the SDMT (-0.79) and paced tapping coefficient of variation (CV) (-0.79) compared to the measures of across-task variability [CV (0.55); intra-individual standard deviation (0.26)]. Across-task variability may be a sensitive marker of cognitive decline in individuals with prodromal HD approaching disease onset. However, individual neuropsychological tasks, including a measure of within-task variability, produced larger effect sizes than an index of across-task IIV in this sample.