This paper draws on a sample of 48,444 scientists and engineers in the United States to analyze nonstandard work arrangements among women and men, 8,773 of whom worked in such arrangements. With few exceptions, women were overrepresented in these arrangements and particularly in those characterized by lower wages and benefits, but their overrepresentation in the worst arrangements failed to explain the gender pay gap. Unlike in the general labor force where “equality at the bottom” tends to prevail, the gender pay gap in science and engineering is greater in the worst nonstandard work arrangements than in the best, and gender equality is greater in the best arrangements than in the worst, possibly because of a selection effect.