Under ideal circumstances fatigue crack growth and fracture data would be generated using large laboratory test specimens with characteristic dimensions that are representative of the final application domain.  Often this means that the preferred test configuration requires a significant investment in material and testing resources.  However, cost, time and material availability may preclude wide panel testing, especially during early trade studies that guide the material selection process.  These cost, time and availability trade-offs can lead to preliminary design curves from much smaller specimens, leading to associated compromises in the form of specimen size effects for the design curves used for trade studies and preliminary design.  Examples of specimen size effects include: (1) fatigue crack growth, where region-III data from smaller specimens typically have higher growth rates than wide panels at the same applied driving force;  (2) spectrum, where fatigue lives of small specimens are not consistent with wide integral structures, and (3) fracture, where smaller specimens fail at much lower toughness values than wide panels.   In each of these cases the small specimens lead to design curves that are unrealistically conservative.  However, advanced simulation and design tools are often able to overcome these limitations, leading to more realistic estimates for preliminary design curves and ultimately to better preliminary designs.  This talk will: (1) contain examples of test data that demonstrate the size effects described above; (2) include analysis results that account for the size effects; (3) present simple corrections where possible to account for size effects; (4) and demonstrate the importance of understanding and accounting for these effects through trade study examples with fatigue crack growth life and residual strength predictions. Finally, this talk will outline areas that standard test methods can be modified to account for or guard against these limitations when testing small specimens.