This paper presents a systematic approach to study the effect of manufacturing variables on the creation of defects and the effect of those defects on the durability of lead-free (Pb-free) solder joints. An experiment was designed to systematically vary the printing and reflow process variables in order to fabricate error-seeded test assemblies. The error-seeded samples were then inspected visually and with X-ray, to identify different types of defects, especially voids, and then test for electrical performance. The specimens were subjected to an accelerated thermal cycling test to characterize the durability of these error-seeded specimens and to study the effect of each manufacturing variable on the durability of the solder joints. The response variables for the design of experiments are thermal cycling durability of the solder joints and void area percentage in ball grid array (BGA) solder joints. Pretest microstructural analysis showed that specimens produced under inadequate reflow profiles suffered from insufficient wetting and insufficient intermetallic formation. Statistical analysis of the response variables shows that waiting time, heating ramp, peak temperature, and cooling rate have nonlinear effects on thermal cycling durability. Two variables in particular [peak temperature and waiting time (the time waited after the solder paste barrel was opened and before print)] appear to have optimum values within the ranges investigated. Statistical analysis of void percentage area for all design of experiment (DOE) runs show that higher stencil thickness results in higher void percentage and that void percentage increases as time above melt and peak temperature increases.