Abstract For two-choice tasks in which stimulus and response locations vary along horizontal and vertical dimensions, the spatial compatibility effect is often stronger on the horizontal than vertical dimension. Umiltà and Nicoletti [(1990) Spatial stimulus-response compatibility (pp. 89–116). Amsterdam: North-Holland] attributed this right-left prevalence effect to an inability to code vertical location when horizontal codes are present simultaneously. Hommel [(1996) Perception & Psychophysics, 43, 102–110] suggested instead that it reflects a voluntary strategy. This study reports four experiments that examine this issue. Experiment 1 was a conceptual replication of Hommel's Experiment 1, with responses made on a numeric keypad and subjects instructed in terms of the vertical or horizontal dimension. The results replicated Hommel's findings that showed a right-left advantage with horizontal instructions; however, with vertical instructions, we found a benefit of vertical compatibility alone that he did not. This benefit for vertical compatibility alone was eliminated in Experiment 2 using a varied practice schedule similar to that used by Hommel. Experiment 3 showed right-left prevalence and a benefit of vertical compatibility alone, even with varied practice and vertical instructions, when subjects responded on perpendicularly arranged handgrips. These benefits were eliminated in Experiment 4 using Hommel's method of urging subjects to respond only in terms of the instructed dimension. With bimanual responses, right-left prevalence is a robust phenomenon that is evident when comparing across vertical and horizontal instructions and, when the right-left distinction is relatively salient, within the vertical instructions condition alone.