This paper examines the integration of geometric and feature cues in spatial memory with a modified version of the spatial reorientation task. Proponents of the geometric module position argue that there is an encapsulated reorientation module, and that true reorientation is accomplished only via geometry (Lee & Spelke, 2010). Features can be used in a second and separate step, but only to locate a goal rather than as a reorientation cue. In contrast, proponents of the adaptive combination approach argue that geometry and feature cues are integrated in spatial memory and that both cues can be used to facilitate reorientation (Newcombe & Ratliff, 2007). The present experiment provides support for the adaptive combination position. The first phase of the experiment was a reorientation task in a trapezoid enclosure (unambiguous geometry) with a narrow feature strip (unambiguous feature). During training, participants started facing a set direction, were disoriented, and then were asked to return to the original facing direction using the geometry, the feature, or both. During these training trials, participants were biased in the direction of the feature, indicating that both features and geometry might be integrated in a Bayesian fashion in spatial memory. The test trials provided additional support for integration and the adaptive combination position. There were four conflict test conditions where the feature moved in relation to the geometry present in training. When the feature moved a small distance in the direction of the geometry, when the feature moved to a novel wall, or when the feature moved a small distance away from the geometry of an acute corner angle, participants selected the correct geometric position. In contrast, when the feature moved a small distance away from the geometry of an obtuse corner angle or when the feature moved a moderate distance away from the geometry of either an acute or an obtuse angle, participants integrated their response and selected and area between the training geometry and feature. The modularity position predicts choices exclusively at the geometry position. This was not found. The adaptive combination position predicts that features and geometry will sometimes be used independently and crucially can be integrated. The data support this position. In sum, the present research provides the first evidence for integration of geometric and feature cues in adult spatial memory during reorientation, and supports the adaptive combination position. Additionally, a mini spatial battery was administered to participants. Correlations of individual differences in spatial skills with reorientation in this paradigm are discussed. / Psychology