CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy) is the most common monogenic cause of stroke and vascular dementia. Accumulation and deposition of the NOTCH3 (N3) extracellular domain in small blood vessels has been recognized as a central pathological feature of the disease. Recent experiments suggested enhanced formation of higher order multimers for mutant N3 compared with wild-type (WT). However, the mechanisms and consequences of N3 multimerization are still poorly understood, in part because of the lack of an appropriate in vitro aggregation assay. We therefore developed and validated a robust assay based on recombinant N3 fragments purified from cell culture supernatants. Using single-molecule analysis techniques such as scanning for intensely fluorescent targets and single-particle fluorescence resonance energy transfer, we show that spontaneous aggregation is limited to CADASIL-mutant N3, recapitulating a central aspect of CADASIL pathology in vitro. N3 aggregation requires no co-factor and is facilitated by sulfhydryl crosslinking. Although WT N3 does not exhibit multimerization itself, it can participate in aggregates of mutant N3. Furthermore, we demonstrate that thrombospondin-2, a known interaction partner of N3, co-aggregates with mutant N3. Sequestration of WT N3 and other proteins into aggregates represents a potentially important disease mechanism. These findings in combination with a new assay for single-molecule aggregation analysis provide novel opportunities for the development of therapeutic strategies.