The trimester-specific associations of maternal dietary patterns with preterm birth (PTB) are unclear. In a prospective prebirth cohort study, we aimed to examine the critical time window of maternal prenatal dietary patterns and the risk of PTB. We assessed prenatal dietary intake among 1500 pregnant women with validated food frequency questionnaires during the 1st, 2nd and 3rd trimester, respectively. We used logistic regression models and generalized estimating equation models to examine the trimester-specific associations and longitudinal associations between maternal dietary patterns in relation to risk of PTB and PTB subtypes. The incidence rate of PTB was 11.9% (179 out of 1500 pregnant women) in the present study. We observed that maternal adherence to a fish-seafood pattern in the 1st trimester was associated with higher risk of PTB [tertile 3 (T3) vs. tertile 1 (T1): OR = 2.29, 95% CI: 1.32-3.96] and iatrogenic preterm birth (IPTB) (T3 vs. T1: OR = 2.26, 95% CI: 1.21-4.20), while a fish-seafood pattern in the 2nd trimester was associated with lower risk of PTB (T3 vs. T1: OR = 0.49, 95% CI: 0.25-0.93). Maternal adherence to a dairy-egg pattern in the 2nd or 3rd trimester was associated with higher risks of PTB and IPTB. No dietary patterns were associated with spontaneous preterm birth. Our findings provide new evidence that specific dietary patterns during different trimesters may have different and even inverse health effects on pregnant women. This supports the necessity of guiding the maternal diet according to different periods of pregnancy to prevent PTB.