Two major measures of negative affectivity (NA) and positive affectivity (PA), developed from the theory of independent affect, were examined. It was found in Study 1 that NA and PA items tended to be judged either very undesirable or desirable when describing people in general. Study 2 revealed that there was a strong positive relation between item desirability and the probability of endorsing the item. Furthermore, the hypothesis that the NA-PA relation varies as a function of respondents' social desirability was generally supported across measures by means of subgroup comparison and moderated regression analysis. Specifically, there was a moderately strong NA-PA relationship only in the group with high social desirability. The relationship was weaker, however, in the group with low social desirability. Potential reasons why NA and PA are related and unrelated under different conditions were discussed. (Copyright 1997, Lawrence Erlbaum.)