In many contexts, decision-making requires an accurate representation of outcome variance—otherwise known as ‘‘risk'' in economics. Conventional economic theory assumes this representation to be perfect, thereby focusing on risk preferences rather than risk perception per se [1–3] (but see [4]). However, humans often misrepresent their physical environment. Perhaps the most striking of such misrepresentations are the many well-known sensory aftereffects, which most commonly involve visual properties, such as color, contrast, size, and motion. For example, viewing downward motion of a waterfall induces the anomalous biased experience of upward motion during subsequent viewing of static rocks to the side [5]. Given that after-effects are pervasive, occurring across a wide range of time horizons[6] and stimulus dimensions (including properties such as face perception [7, 8], gender [9], and numerosity [10]), and that some evidence exists that neurons show adaptation to variance in the sole visual feature of motion [11], we were interested in assessing whether after-effects distort variance perception in humans. We found that perceived variance is decreased after prolonged exposure to high variance and increased after exposure to low variance within a number of different visual representations of variance. We demonstrate these after-effects occur across very different visual representations of variance, suggesting that these effects are not sensory, but operate at a high (cognitive) level of information processing. These results suggest, therefore, that variance constitutes an independent cognitive property and that prolonged exposure to extreme variance distorts risk perception—a fundamental challenge for economic theory and practice.