Sign detection is a valuable part of intelligent driving aids and of the developments towards selfdriving or autonomous vehicles. The impact of various environmental conditions on such systems needs to be explored and this study aims to fill a gap in current research relating to the effect of sign rotation on such systems. This is approached in a quasi-experiment varying only the rotation of sign images and measuring the resulting accuracy. The study finds that upside-down signs are significantly more difficult to detect than right-side-up signs and that sideways-rotated signs are far more difficult than that. The conclusion drawn is that resources likely need to be spent towards negating this limitation when similar systems are developed to ensure high performance even in cases where signs have been rotated.