We investigated predictors of change in plasma carotenoids from baseline to 3 y and examined plasma carotenoid concentrations at 1 and 3 y in response to a high vegetable diet. Participants were 56 women diagnosed with breast cancer and enrolled in a randomized feasibility study for a trial examining the effect of a diet high in vegetables and fruits on the risk of breast cancer recurrence. Independent t test analysis revealed that the intervention group had significantly higher vegetable and fruit servings and fiber at 12 mo and significantly higher vegetable servings at 36 mo compared with the control group (P < 0.05). Energy intake from fat was significantly lower in the intervention group at 12 and 36 mo. The intervention group had significantly higher consumption of beta-carotene, alpha-carotene, lutein and beta-cryptoxanthin at 12 mo (P < 0.05). beta-Carotene, alpha-carotene and lutein intakes also were significantly higher at 36 mo (P < 0.05). At 36 mo, the intervention group had significantly higher plasma concentrations of alpha-carotene and beta-carotene compared with the control group. Repeated-measures ANOVA revealed that the intervention group had significantly increased (P < 0.05 with Bonferroni correction) plasma beta-carotene, alpha-carotene, lutein and lycopene concentrations at 12 and 36 mo compared with baseline. Baseline carotenoid concentrations were significantly inversely predictive (P < 0.05) of plasma carotenoid change. In addition, change in body mass index (BMI) and plasma cholesterol concentrations were predictive of plasma carotenoid change from baseline to 3 y. Results of this study demonstrate that change in plasma carotenoid concentrations is associated with change in BMI, change in plasma cholesterol and baseline carotenoid concentrations. Plasma carotenoid response can be an indicator of long-term high vegetable intake for women at risk of breast cancer recurrence.