Predicting the fatigue life of solder interconnections is a challenge due to the complex nonlinear behavior of solder alloys, the importance of the load history, and the wide variation in package constructions employed by the microelectronics industry. While material behavior can be captured by means of valid constitutive models, the empirical failure models in common use are strongly dependent on package construction due to the fact that they are derived from tests of solder joints with heterogeneous stress and microstructural variation. Despite this limitation, the empirical models have been used to identify reliable choices among package design alternatives. In this technical brief, some of the key limitations that the user of empirical models must be aware of are addressed. Physical arguments are proposed in order to justify the appropriate assumptions that may be made in the life model. Using data from two different sources in the literature, the constants in the empirical models are shown to be dependent on the solder joint geometry. Alternative fits to experimental data that reduce the dependence of the fits on solder joint geometry are proposed.