Four experiments with adults on Piaget's (J. Piaget & B. Inhelder, 1956) water-level task investigated previous reports that containers tilted at larger angles produce more error than those tilted at smaller angles. Experiment 1 found that this orientation effect occurred in both male and female subjects when the angles were manipulated between subjects but was absent in male subjects when manipulated within subjects. Experiments 2 and 3 indicated that male, and to a lesser degree female subjects, appear to benefit from exposure to easier (less-tilted) stimulus trials and, as a result, perform more accurately on subsequent difficult trials. Experiment 4 implicated mental rotation processes in the orientation effect by demonstrating that accuracy varied as a function of how far subjects needed to mentally rotate the container from its initial position to its test position. A developmental sequence is proposed in which relevant spatial abilities are assumed to facilitate acquisition of the principle that liquids are invariantly horizontal, which in turn leads to more accurate performance on the task. Differences in cognitive strategies may be responsible for the Gender X Orientation interaction observed under some conditions. One of Piaget's most fascinating developmental tasks is the water-level problem, designed originally to assess children's emerging understanding of Euclidean (three-dimensional) space (Piaget & Inhelder, 1956). According to Piaget, at some point during the concrete operations period, children come to realize that the location of objects can be specified in relation to stable vertical and horizontal coordinates in the environment. The development of this spatial system, in turn, leads children to observe in their everyday activities that the surface of a liquid remains invariantly horizontal regardless of the orientation of its container. Thus, when Piaget presented subjects with a partially filled bottle and asked them to predict how the water line would look if the bottle were tilted, he reported that those beyond the age of 9 years generally indicated that the surface would remain horizontal. It has become increasingly clear, however, that the water-level task may not be a very good index of this kind of spatial knowledge (Liben, 1991). Although children in the concrete operations period do indeed develop an understanding of Euclidean space—as demonstrated by their performance on other spatial tasks—there is little evidence that this cognitive accomplishment results directly in their acquiring the principle that liquids remain invariantly horizontal. In fact, subsequent research with the water-level problem has instead found that many adolescents and adults perform poorly on the task and also cannot