Previous research has demonstrated that grammatical gender in Dutch is typically acquired late. Most of this work used production data only, and consequently children’s knowledge of Dutch gender may have been underestimated. In this study, therefore, we examined whether 49 4- to 7-year-old Dutch-speaking children (and 19 adult controls) were able to use gender marking in the article preceding the object label during online sentence processing to (a) anticipate the upcoming object label or to (b) facilitate the processing of that label as it is presented. In addition, we investigated whether children’s online processing and production of gender marking on articles were related. In an eye-tracking task, participants were presented with sentences and visual displays with two objects, representing nouns of either the same gender (uninformative) or different genders (informative). Children were divided into a non-targetlike group and a targetlike group on the basis of their scores for neuter nouns in the production task. Our analyses examined whether participants could use gender marking anticipatorily (i.e., before the onset of the noun) and facilitatively (i.e., from noun onset). Results showed that Dutch-speaking adults and children who were successful in production used gender marking anticipatorily. However, children who did not systematically produce gender-marked articles used gender marking only facilitatively. These findings reveal that successful online comprehension may in part be possible before targetlike production is completely in place, but at the same time targetlike production may be a trigger for online comprehension to be completely successful.