This paper investigates the role of parental input and transparency in the acquisition of two different gender systems, Norwegian and Russian, by bilingual children living in Norway. While gender in Russian is generally predictable from the morphophonological shape of the noun (with some exceptions), gender assignment in Norwegian is opaque. An experimental production study was carried out with two groups of bilinguals, children with one or two Russian-speaking parents, and monolingual controls (age 4;1–7;11). The findings show that both groups of bilinguals perform similarly to monolinguals in Norwegian, the majority language, despite the lack of transparency. In Russian, on the other hand, not only quantitative, but also qualitative differences are found in the data of the bilingual children with the least exposure to the language. These qualitative differences indicate that early age of onset is not sufficient to acquire phenomena such as gender; extensive input is necessary.