Organic semiconductors have recently emerged as a promising alternative to standard inorganic semiconductors for many commercial applications. Their ability to be processed from solution combined with their unique mechanical resistance paves the way for low cost roll-to-roll production of flexible devices. While, for screen and lighting applications, organic light emitting diodes are already well-established on the market, further understanding and development remains to be achieved for organic-based field effect transistor and photovoltaic devices to be commercially viable. Especially, the impact of supramolecular self-assembly on the opto-electronic properties of organic semiconductors persist in being a key challenge. The original work presented in this thesis explores the misunderstood relationship between supramolecular assembly of solution processed organic semiconductors and the corresponding performances in transistor and photovoltaic applications. Additionally, novel strategies and tailored molecular tools are developed as to permit rational engineering of the self-assembly and offer unprecedented control over the device performances. In brief, this thesis first addresses the controversial band-like temperature dependence of the charge carrier mobility in organic semiconductors. An alternative physical origin for this behaviour is proposed by considering a temperature dependent biphasic equilibrium in solid state using a small molecule, coded DPP(TBFu)2, which offers a unique case study for this new model. This work then explores the underlying mechanism behind aliphatically-linked organic semiconductors to control the crystal structure, focusing on two distinct dimers based on DPP(TBFu)2. The linker position in these dimers is found to greatly affect the self-assembly and the driving force governing crystallization events, overall leading to either enhanced hole mobility in transistor or enhanced thermal stability in photovoltaic devices. Following work demonstrate a simple charge cascade self-assembly in solar cells based on crystallinity differences between a host blend and an amorphous ternary compound, exceeding 30 % enhancement in the power conversion efficiency. Finally, the effect of the polymer molecular weight on blend self-assembly and photovoltaic performances is investigated for an amorphous polymer coded PDPP4T-TT and a semicrystalline one, coded PBTTT.