To elucidate the mechanism by which pre-aggregation and miscibility matching govern the active layer morphology in non-fullerene organic solar cells, chloroform (CF) and o-xylene (OX) were used as solvents, while D18 and N2200 were incorporated as third components into the PM6:L8-BO system. The incorporation of D18 enhanced device performance, whereas the addition of N2200 reduced device performance. Based on surface energy analysis, the free energies of pure components and binary blends in different solvents were calculated, showing that the Gibbs free energies of D18, PM6 and L8-BO exhibited better pre-aggregation matching. Employing the melting point depression method, the Flory–Huggins interaction parameters of D18?:?L8-BO (1?:?6) and N2200?:?L8-BO (1?:?5) blends were calculated. The results revealed that the miscibility of the samples cast with CF was superior to those cast with OX. Atomic force microscopy (AFM) and transmission electron microscopy (TEM) observations revealed that D18 could induce L8-BO to aggregate and crystallize to form a nanofiber architecture, leading to an optimized phase separation. Attributing to the desirable miscibility of D18 with L8-BO, the D18:L8-BO nanostructures could be dispersed within an amorphous PM6 matrix, forming a double-fibril network morphology that facilitated charge transfer and enhanced device performance. In contrast, N2200 was immiscible with L8-BO, which led to the formation of a suboptimal morphology exhibiting excessive aggregation or excessive dispersion, resulting in a deterioration in charge transfer and device performance. The investigation of pre-aggregation matching in solvents, and miscibility matching of the components could provide guidance for the rational selection of appropriate solvents and suitable third components towards high-performance ternary organic solar cells.