In the field of Human-Computer Interaction there is considerable awareness on diversity and inclusion. At the same time topics such as gender and race have become more prominent recently. One aspect that has received little attention, however, is the possible reproduction of real-world socio-demographic inequality structures through recommendation systems in fashion. To investigate gender-specific differences in recommender systems, we utilise data from Amazon and use quantile regressions to calculate what price differences exist for the recommended products concerning the primary product. Our results show a bias in recommended pricing premiums about addressed gender. While a higher price in comparison to the viewed product is charged for all genders, product recommendations for women generally show a higher premium than those for men (about 5% more at the median, ceteris paribus). This can be influenced by the starting price and the popularity of the product, i.e. the sales ranking.