This paper looks at online representations of women and men in science, engineering and technology. We show that these representations largely re/produce dominant gender discourses. We then focus on the question: How are gender cliched images re/produced online? Drawing on a discursive analysis of data from six interviews with web authors, we argue that there are two reasons why their awareness of gender issues does not always translate into website content. First, web authors think of themselves as working within either journalistic or scientific cultures, and draw on associated criteria, which exclude gender equity, to make content decisions. Second, they construct distinctions between representation and reality, judging representations on their "empirical realism" (how accurately they represent a reality seen to exist outside the text), foreclosing considerations of their productive power. Finally, we draw on interview data with young people to show that these constrain the meanings made by web users. (Contains 3 notes.)