Background: Evidence suggests that somatic rather than cognitive depressive symptoms are risk factors for recurrent cardiac events in at-risk patients. However, this has never been explored using a time-dependent approach in a narrow time-frame, allowing a cardiac event-free time-window. Methods: The analysis was performed on 595 participants [70.6% male, median age 72 (27–98)] drawn from the UPBEAT-UK heart disease patient cohort with 6-monthly follow-ups over 3 years. Depressive symptomatology was measured using the Patient Health Questionnaire-9 (PHQ-9) (four somatic, five cognitive items). New cardiac events (NCEs) including cardiac-related mortality were identified by expert examination of patient records. Analyses were performed using Cox proportional hazard models with delayed entry, with time-dependent depressive dimensions and covariates measured 12–18 months (median: 14.1, IQR: 3.5) prior to the event, with a 12-month cardiac event-free gap. Results: There were 95 NCEs during the follow-up [median time-to-event from baseline: 22.3 months (IQR: 13.4)]. Both the somatic (HR 1.12, 95% CI 1.05–1.20, p = 0.001) and cognitive dimensions (HR 1.11, 95% CI 1.03–1.18, p = 0.004) were time-dependent risk factors for an NCE in the multi-adjusted models. Specific symptoms (poor appetite/overeating for the somatic dimension, hopelessness and feeling like a failure for the cognitive dimension) were also significantly associated. Conclusion: This is the first study of the association between depressive symptom dimensions and NCEs in at-risk patients using a time-to-event standardised approach. Both dimensions considered apart were independent predictors of an NCE, along with specific items, suggesting regular assessments and tailored interventions targeting specific depressive symptoms may help to prevent NCEs in at-risk populations.