Sixty-six rats of seven inbred strains were observed in an open field (OF) and in a light-dark preference box (LDB). Each set of apparatus provides measures of locomotor activity and of the time spent in the "most protected" area (the corners in the OF, the dark compartment in the LDB). These measures are proposed to be indices of emotionality. In the present study, the reliability and validity of the OF and LDB measures were determined by genetic correlations. Genetic correlations (rGEN) within and between tests were used as they provide a more coherent picture of reliability and validity of the measures than product-moment correlations (rPM), since rGEN's are appropriate estimates of "true" individual scores. It appeared that the activity measure in the OF and the occupancies of the most protected parts of OF and LDB were valid indices of emotionality, while the activity measure in the LDB was not. We recommend that the occupancy measures in the OF and the LDB should be used in addition to the commonly used activity measure in the OF to assess emotionality. The results indicate that genetic correlations provide a useful tool to validate measures in animal research. The only experimental manipulation in this approach consists of using the variation between inbred strains.