Evidence suggests that enclosure size influences the relative use of local but not global geometric cues during spatial reorientation (Sturz, Forloines, and Bodily, 2012). Yet, it remains unclear whether training size, testing size, or the relationship between training size and testing size is responsible for the relative use of local geometric cues. We examined the extent to which manipulations of enclosure size during training (while keeping testing size constant) or at testing (while keeping training size constant) influenced the relative reliance on corner angles (a local geometric cue) during reorientation. Such a design also allowed us to investigate the potential influence of the relationship between training and testing size (i.e., increase in size from training to testing or decrease in size from training to testing). We trained participants to respond to a unique goal location in differently sized trapezoidal enclosures; we then tested participants in a rectangular enclosure to isolate the use of a global geometric cue and a parallelogram to place local and global geometric cues in conflict. Results suggest that the relationship between training and testing size influenced the relative use of local but not global geometric cues for spatial reorientation.