Although an overwhelming proportion of all legal disputes end in settlement, the determinants of the timing of settlement remain empirically underexplored. We draw on a novel dataset on the duration of commercial disputes in Slovenia to study how the timing of settlement is shaped by the stages and features of the litigation process. Using competing risk regression analysis, we find that events such as court-annexed mediation and the first court session, which enable the disputing parties to refine their respective expectations about the case outcome, in general reduce case duration to settlement. The magnitude of the respective effects, however, varies with time. Completion of subsequent court sessions, in contrast, does not affect the time to settlement. Judicial workload affects the timing of settlement indirectly, via the effect on the timing of the first court session. We also examine the effect of other case and party characteristics.