The benefits and value of new cancer treatments often focus on the overall survival (OS) gains that patients may derive. Trials are typically not long enough to allow detailed understanding of OS, and potential benefits must be inferred from benefits on progression-free-survival (PFS). This raises questions such as whether early or later progression impacts survival, whether the increase in mortality following progression is sustained or gradually diffused, and whether a benefit observed on PFS implies a benefit in OS. Answering these questions requires an analytical framework in which progression and survival can be analyzed together and parameterized to address key questions. We propose a statistical modeling framework based on Cox regression and time-dependent predictors and effects. A simple formulation of this model would include a time-dependent indicator for progression, whose coefficient would measure the increase in risk of death following the event. This is very limiting, however; it assumes that the timing of progression does not matter and that the increase in risk of death is sustained indefinitely. A more flexible formulation can be built using two descriptors of event: the timing of progression (TP) and time since progression (TSP). These can be continuous measures or categorized (e.g., early vs. late TP), as appropriate. The coefficient for TP reveals whether later progression is associated with higher/lower subsequent mortality, while the coefficient of TSP reflects whether and for how long the increase/decrease in mortality is sustained and whether it ever returns to the level of patients who had not progressed. The impact of treatment can be captured on each of these parameters separately. The proposed framework will be illustrated with an example, and extension of the approach to other applications (e.g., measuring the impact of a stroke on survival) will be discussed.