Objective This study employed a longitudinal design to investigate perinatal post-traumatic stress disorder in high-risk pregnant women across multiple time points, aiming to identify optimal timing for intervention. Methods A longitudinal research design was used to assess the P-PTSD inventory scores at 1 month after diagnosis of high-risk pregnancy (T 0 ), 3 months after diagnosis (T 1 ), 6 months after diagnosis (T 2 ), and 1 month postpartum (T 3 ), to analyse the overall trend of P-PTSD using generalized estimating equations, and to determine the heterogeneity of P-PTSD in different characteristics of high-risk pregnancies by using a group trajectory model. Results Among 245 high-risk pregnant women, the positive detection rates of P-PTSD at T0–T3 were 12.7, 29.8, 31.4, and 35.9%. Generalized estimating equations revealed a significant overall upward trend (χ 2 = 64.02, p < 0.001). Group-based trajectory modeling identified three distinct trajectory classes—low-risk stable (36.1%), moderate-risk worsening (46.8%), and high-risk fluctuating (17.1%)—with the 3-group solution (polynomial order 1, 2, 2) selected based on the lowest BIC (−3438.08) and acceptable classification accuracy (Entropy = 0.71). Multinomial logistic regression identified several independent predictors of trajectory membership, including monthly household income (e.g., 6,001–9,000 yuan vs. lower income: OR = 0.19, 95% CI: 0.04–0.86) and prior trauma history (no vs. yes: OR = 0.17, 95% CI: 0.03–0.93). Conclusion This study found through longitudinal tracking that P-PTSD symptoms in high-risk pregnant women generally worsened over time, and that there was significant heterogeneity in the development trajectory of symptoms in this group. Socioeconomic status, perinatal-related factors, and trauma experiences are the core predictive variables distinguishing different symptom development trajectories.