Asymmetry of paw usage in the laboratory mouse is an experimental model for left-right asymmetry of hand usage. Given a set number of reaches into a centrally placed food tube (an unbiased or U-world test), individual mice exhibit a number of left and right paw reaches that is reliably expressed on retesting. Whereas different inbred strains appear to have equal numbers of individual mice with a left- or a right-preferred paw after a U-world test, there are significant differences among strains in the degree or strength of lateralization of the preferred paw. We report here a systematic series of tests of paw usage with naive mice and retests of the individuals in test chambers with the food tube biased to the left or to the right, contrasting the highly lateralized C57BL/6J and the very weakly lateralized (or ambilateral) CDS/Lay inbred strains and their (B6 × CDS) F1 generation. The results caused a shift in the paradigm of paw usage. There is an unexpected qualitative difference in paw usage between C57BL/6J and CDS/Lay. C57BL/6J is random in its left-right paw usage, but it is conditioned by the left or right direction of the initial biased-world test and by usage. CDS/Lay is constitutively equal-pawed, responds very little to direction of the test chamber, and is not conditioned by it. The probability of left-paw versus right-paw usage depends on both the genotype and the context of the test. The (B6 × CDS) F1 generation suggests that constitutive equal-paw usage of CDS/Lay is dominant to experience-conditioned paw usage of C57BL/6J. There is also an apparent quantitative difference between the very weakly lateralized (ambilateral) preferred paw usage in CDS/Lay and the highly lateralized preferred paw usage in C57BL/6J. The difference in degree of lateralization of preferred paw usage between the constitutively equal-pawed CDS/Lay strain and (B6 × CDS) F1 generation must originate from allelic differences at other gene loci between the CDS/Lay and C57BL/6J parental strains. The SWV and NOD/Lt strains were also assessed in asymmetrical tests because they were known to be weakly lateralized and similar to each other in a U-world test and to be significantly different from both C57BL/6J and CDS/Lay. SWV is experience-conditioned and weakly lateralized; NOD/Lt is constitutively equal-pawed and weakly lateralized. Further analysis will determine the genetic cause of the qualitative difference between constitutive equal-paw and experience-conditioned paw usage and the genetic cause of the quantitative differences in degree of lateralization of the preferred paw within each type of paw usage.Key words: mouse, left-right handedness, behavioural genetics, phenotypic reaction norms, constitutive behaviour, experience-conditioned behaviour, memory.