Metastasis is the main cause of cancer-related deaths, but efficient targeted therapies against metastasis are still missing. Major gaps exist in our understanding of the metastatic cascade, as existing methods cannot combine sensitivity, robustness, and practicality to dissect cancer progression. Addressing this issue requires improved strategies to distinguish early metastatic colonization from metastatic outgrowth. Luciferase-labelled MDA-MB-231, MCF7, and 4T1 breast cancer cells were spiked into samples from tumour-naïve mice to establish the limit of detection for disseminated tumour cells. Luciferase-labelled breast cancer cells (±unlabelled cancer-associated fibroblasts; CAFs) were orthotopically implanted in immunocompromised mice. An ex vivo luciferase assay was used to quantify tumour cell dissemination. In vitro luciferase assay confirmed a linear and positive correlation between cancer cell numbers and the bioluminescence detected at single cell level in blood, brain, lung, liver, and mammary fat pad samples. Remarkably, single luciferase-labelled cancer cells were detectable in all of these sites, as the bioluminescence quantified in the analysed samples was substantially higher than background levels. Ex vivo, circulating tumour cells, metastasis, and tumour self-seeding were detected in all samples from animals implanted with highly metastatic luciferase-labelled MDA-MB-231 cells. In turn, detection of poorly metastatic luciferase-labelled MCF7 cells was scarce but significantly enhanced upon co-implantation with CAFs as early as 20 days after the experiment was initiated. These results demonstrate the feasibility of using an ultrasensitive luciferase-based method to dissect the mechanisms of early metastatic colonization to improving the development of antimetastatic therapies.