An experiment is reported comparing human performance on two kinds of visually presented traveling salesperson problems (TSPs), those reliant on Euclidean geometry and those reliant on city block geometry. Across multiple array sizes, human performance was near-optimal in both geometries, but was slightly better in the Euclidean format. Even so, human solutions varied systematically across geometries, with the differences reflecting sensible adaptations to the different distance structures that these geometries imply. These results suggest that human near-optimality for visually presented TSPs may apply to a more general class of problems than the standard Euclidean case.