Abstract: Background: No accurate method, clinical or otherwise, currently exists to determine the onset of labor precisely. The objective of this study was to investigate what influences the duration of first stage labor in women with spontaneous labor and childbirth in a nonclinical setting. Methods: From a population-based cohort of 1,448 planned home and birth center births, we selected 932 births for absence of pathology, absence of intervention, and completeness of data. Duration of first stage labor was analyzed with regression analysis for duration data or time-to-event analysis, using a specialized Transition Data Analysis software. The effects of fixed (age, parity, education, antenatal classes, infant birthweight, first cervical assessment) and time-varying factors (start of midwifery care, spontaneous rupture of membranes) in labor were estimated with piecewise-constant exponential hazard models. Results: Of the characteristics immutable at the onset of labor, only parity had a strong effect on the duration of first stage labor. Cervical dilatation at first assessment and time-varying factors, such as the timing of spontaneous rupture of membranes and midwifery care, each had a strong influence on labor duration; however, the sequence in which they occurred exerted an even stronger influence. First stage labors were much shorter if the membranes ruptured before rather than after the start of care. Conclusion: With the exception of parity, events occurring during labor and their timing have a greater influence on the duration of first stage spontaneous labor than elements which are immutable at the onset of labor. Trials of interventions to influence the duration of labor need to consider not only whether the intervention was applied or not, but also when it was applied, if cause-effect relationships are to become properly understood.