In Polish, where different inflectional paradigms apply to female and male names, attributing male gender to female referents such as eminent scientists and authors that pass unnoticed in English becomes visible in scholarly as well as popular scientific texts, exposing the society's gender bias. While gender bias in machine translation has been widely observed and criticised, it is believed that human translation would eliminate factual mistakes of this type. Yet false attributions of male gender to eminent female linguists found in Polish scholarly texts suggest the contrary. A translation experiment conducted with a group of graduate students of translation was conducted in order to test the presence and strength of the bias, differences between male and female student translators, and the impact of earlier teaching on assigning gender in translation from English to Polish. Both male and female students frequently attributed the male gender to linguistic scholars even in the presence of strong evidence to the contrary, and after prior learning. Male students' choices were slightly but significantly more biased. The conclusions pertain to the social consequences of the bias and the need to include this topic in translation training while attending more strongly to students' research and sociocultural competence.