The use of blood-circulating cell-free DNA (cfDNA) as 'liquid-biopsy' is explored worldwide, with hopes for its potential in providing prognostic or predictive information in cancer treatment. In exploring cfDNA, valuable repositories are biobanks containing material collected over time, however these retrospective cohorts have restrictive resources. In this study, we aimed to detect tumor-specific mutations in only minute amounts of serum-derived cfDNA by using a targeted next generation sequencing (NGS) approach. In a retrospective cohort of ten metastatic breast cancer patients, we profiled DNA from primary tumor tissue (frozen), tumor-adjacent normal tissue (formalin-fixed paraffin embedded), and three consecutive serum samples (frozen). Our presented workflow includes comparisons with matched normal DNA or in silico reference DNA to discriminate germline from somatic variants, validation of variants through the detection in at least two DNA samples of an individual, and the use of public databases on variants. By our workflow, we were able to detect a total of four variants traceable as circulating tumor DNA (ctDNA) in the sera of three of the ten patients.