Background: Up to 30% of patients with breast cancer relapse after primary treatment. There are no sensitive or reliable tests to monitor these patients and detect distant metastases before overt recurrence. Here, we demonstrate the use of personalized circulating tumour DNA (ctDNA) profiling performed postoperatively, postadjuvantly, and serially for detection of recurrence in breast cancer. Methods: Patients with primary breast cancer (n=188) were recruited following surgery and adjuvant therapy and were followed-up for up to 10 years with semi-annual blood sampling for ctDNA analysis. Patients (n=29) with insufficient residual tumour for whole exome sequencing (WES) were excluded from the analysis. Tumour WES profiles were generated for 159 patients; samples from 2 patients failed WES QC requirements and a personalised ctDNA panel could not be generated for 1 patient. In 156 patients, plasma samples (n=1141) were retrospectively tested for the presence of ctDNA using personalized Signatera assays (mPCR-NGS) targeting up to 16 somatic single nucleotide variants selected from primary tumour WES. Results: Plasma ctDNA was detected ahead of clinical or radiologic relapse in 30 of the 34 relapsed patients (sensitivity of 88%). Metastatic relapse was predicted with a lead interval of up to 2 years (median: 10 months, range: 0-39 months); median lead intervals for HR+/HER2- were 15 (2 - 39); for HR-/HER2+ 6 (0.5 – 12) for HR+/HER2+ 8 (5 – 14) and 9 (0-20) for TNBC. Patients with a positive ctDNA test had poorer relapse-free-survival (RFS) (HR=47.5; 95% CI 18.5-161.4; p <0.001) from surgery and all four breast cancer subgroups showed a similarly reduced RFS. Overall survival was also significantly reduced for patients who were ctDNA positive (HR=84.15; 95%CI 16.43-1538; p <0.001). The number of variants, mean VAF and MTM/mL varied between patients, with significantly higher values at the time closest to relapse than in the first ctDNA positive sample (p =0.0002). Among the 4 relapsed patients not detected in the study all were HR+/HER2-, 1 had a local recurrence, 2 had bone recurrence (1 with axillary LN involvement) and 1 had cancer cells in pleural fluid. Of the remaining 122 patients, only 5 developed ctDNA-positivity, all with low VAF, none of them have relapsed by the follow-up census date (31 December 2021). However, follow-up for some of these patients limits definitive assessment. Lastly, 4 patients developed a second primary cancer (2 breast, 2 lung) all of whom were ctDNA-negative. Conclusions: This study demonstrates that serial post-operative ctDNA analysis has strong prognostic value. More importantly, earlier detection of metastatic disease provides a possible window for therapeutic intervention, while repeated negative ctDNA tests can provide reassurance to patients. Future interventional studies may assess the clinical utility of ctDNA-based risk-stratification.