A number of studies have proposed the existence of a distorted body model of the hand. Supporting this hypothesis, judgments of the location of hand landmarks without vision are characterized by consistent distortions—wider knuckle and shorter finger lengths. We examined an alternative hypothesis in which these biases are caused by domain-general mechanisms, in which participants overestimate the distance between consecutive localization judgments that are spatially close. To do so, we examined performance on a landmark localization task with the hand (Experiments 1–3) using a lag-1 analysis. We replicated the widened knuckle judgments in previous studies. Using the lag-1 analysis, we found evidence for a constant overestimation bias along the mediolateral hand axis, such that consecutive stimuli were perceived as farther apart when they were closer (e.g., index-middle knuckle) versus farther (index-pinky) in space. Controlling for this bias, we found no evidence for a distorted body model along the mediolateral hand axis. To examine whether similar widening biases could be found with noncorporeal stimuli, we asked participants to localize remembered dots on a hand-like array (Experiments 4–5). Mean localization judgments were wider than actual along the primary array axis, similar to previous work with hands. As with proprioceptively defined stimuli, we found that this widening was primarily due to a constant overestimation bias. These results provide substantial evidence against a distorted body model hypothesis and support a domain-general model in which responses are biased away from the uncertainty distribution of the previous trial, leading to a constant overestimation bias.