Incorporating particulate metal-oxide nano-fillers into brittle epoxy matrices effectively modifies fracture toughness; however, structural endurance is fundamentally governed by the competition between effective nanoparticle dispersion and localized agglomeration. This study investigates the micro-mechanical behavior, toughening mechanisms, and agglomeration effects of TiO_2 nanoparticles incorporated into a structural epoxy adhesive at 0.50, 1.0, and 3.0 wt.% loading fractions. Glass/epoxy double-lap joints were fabricated and subjected to quasi-static tensile and tension-tension cyclic fatigue testing (R=0.1, 4 Hz) to evaluate these particulate systems. The results indicate that optimal dispersion at 1.0 wt.% yields the best average performance, achieving the highest mean static failure load (8.44 kN) and maximum characteristic fatigue life (\eta=37,857 cycles). Micro-mechanical analysis suggests this optimal fraction enhances fatigue resistance through energy-dissipating mechanisms, predominantly crack pinning and deflection. Conversely, 3.0 wt.% loading severely degrades static and fatigue behavior which is attributed to significant nanoparticle agglomeration, forming rigid clusters that act as localized stress concentrators and promote premature microcrack coalescence. Since direct microscopic characterization of the filler dispersion was not performed, these toughening and agglomeration mechanisms are inferred from the mechanical responses and fracture behavior. Furthermore, Weibull analysis exposed a crucial reliability crossover linked to filler distribution: while 1.0 wt.% exhibited the highest average fatigue life, the more uniform distribution at 0.5 wt.% minimized statistical scatter. This rendered it the safest conservative formulation at a 90% reliability level (B_{10} life). These findings underscore that harnessing the full potential of particulate fillers requires balancing active micro-mechanical toughening with conservative statistical reliability.