Stöber's Social Desirability Scale (SDS-17) was examined psychometrically in 5 samples (N=2817) from Austria, Canada, and the U.S.A. Rasch and Mokken scaling analyses attested the SDS-17 is not strictly unidimensional. Age, agreeableness, and conscientiousness were notable positive correlates of SDS-17 scores. There were signs of non-normal score distributions, acquiescence bias, and sex and country differences (higher scores among Austrians than North Americans). Items with higher ratings of social desirability according to previous research were particularly prone to show sex effects. The SDS-17 appears suitable in cross-cultural settings, but may benefit from substituting its true-false response format with a rating-scale format. A formative-indicators view regarding the social desirability construct and the SDS-17 is discussed.