In this paper, a correspondence testing experiment is conducted to examine sexual orientation discrimination against lesbians in Germany. Applications for four fictional female characters are sent out in response to job advertisements: a heterosexual single, a married heterosexual, a single lesbian and a lesbian who is in a 'same-sex registered partnership'. Different results are obtained for the two cities investigated, Munich and Berlin. While single lesbians and lesbians in a registered partnership are equally discriminated in comparison to the heterosexual women in the city of Munich, no discrimination based on sexual orientation has been found in Berlin. Furthermore, for a subset of our data we can compare the effects of a randomized versus a paired testing approach, which suggests that under certain conditions, due to increased conspicuity, the paired testing approach may lead to biased results.