Objective To explore the distinct longitudinal trajectories of illness perceptions among stroke patients from the acute stage to the convalescence stage and to analyze the determinants of these trajectories. Methods A total of 192 stroke patients were surveyed by utilizing the general information questionnaire, Brief Illness Perception Questionnaire (BIPQ), Herth Hope Scale, Hospital Anxiety and Depression Scale during hospitalization.They were invited to complete the BIPQ at 1, 3 and 6 months post baseline survey. The difference in illness perception at different time points was analyzed, growth mixture model was employed to identify patterns of illness perception change, and decision tree analysis was performed to explore the core determinants of these trajectories. Results The total BIPQ score at 4 time points was 45.16±15.46, 40.30±15.43, 35.07±15.32 and 27.38±14.56 respectively, showing significant difference (P＜0.01). Two clusters of the trajectories of illness perception were identified, and designated as ′average illness perception and rapid decline′ (83.9%) and ′high illness perception and slow decline′ (16.1%).The decision tree model showed that hope, the number of stroke attacks, depression, anxiety, occupational status and age could predict the trajectories.Hope was of paramount importance for it was the root node of the decision tree. Conclusion Illness perception of stroke patients declines post stroke, and shows group heterogeneity. Hope is a core predictor. Patients with high illness perception and slow decline trajectory should be identified as early as possible based on the predictors, and hope may be an additional target of treatment, thus to reduce their illness perception.