A database of mouse locomotor path in spatial tests can be used to search in silico for behavioral measures that better discriminate between genotypes and are more replicable across laboratories. In this study, software for the exploration of exploration (SEE) was used to search a large database for a novel behavioral measure that would characterize complex movement paths. The database included mouse open-field behavior assessed in 3 laboratories, 7 inbred strains, several pharmacological treatments and hundreds of animals. The new behavioral measure, "path texture", was characterized using the local curvature of the path (the change of direction per unit distance, in degrees/cm) across several spatial scales, starting from scales smaller than the animal's body length and up to the scale of the arena size. Path texture analysis differs from fractal dimension analysis in that it does not assume self-similarity across scales. Path texture was found to discriminate inbred strains with relatively high broad-sense heritability (43%-71%) and high replicability across laboratories. Even genotypes that had similar path curvatures in some scales usually differed in other scales, and self-similarity across scales was not displayed by all genotypes. Amphetamine decreased the path curvature of C57BL/6 mice in small and medium scales, while having no effect on DBA/2J mice. Diazepam dose-dependently decreased the curvature of C57BL/6 mice across all scales, while 2 anxiogenic drugs, FG-7142 and pentylenetetrazole, increased it. Path texture thus has high potential for behavioral phenotyping and the study of drug effects in the mouse.