This study investigated the association between dietary variables and plasma levels of perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) among 1076 pregnant women. Diet was assessed at midpregnancy by a food-frequency questionnaire. Mean first trimester plasma PFOS and PFOA levels were 35.1 and 5.6 ng/mL respectively. PFOS levels were positively associated (p < 0.05) with intake of red meat, animal fats, and snacks (e.g., popcorn, potato chips), whereas intake of vegetables and poultry was inversely associated. The adjusted mean differences between the 75th and 25th intake percentiles were 4.3 ng/mL [95% CI: 2.1, 6.5] for red meat 3.4 ng/mL [95% CI: 1.2, 5.6] for animal fats, and 2.0 ng/mL [95% CI: 0.3, 3.6] for snacks. Similar but weaker associations were observed for PFOA. Furthermore, a comparison between women reporting low (< or =25th percentile) red meat and high (> or =75th percentile) vegetable intake and women reporting low vegetable and high red meat intake resulted in differences in plasma PFOS and PFOA concentrations equal to 31% and 18% of mean levels, respectively. Studies quantifying levels of perfluorinated compounds in food have suggested that diet could be an important route of human exposure. The observed associations in our study between dietary variables and maternal exposure further support that conclusion.