Large crabgrass (Digitaria sanguinalis) is one of the most problematic weeds in summer crops in Argentina. Emergence throughout the season of several cohorts allows the weed to escape postemergence control. Demographic models are useful tools to understand and compare the effect of different agronomic management decisions on weed population growth, as well as to identify critical functional stages that affect population growth rates. The objectives of this work were (i) to study population dynamics of D. sanguinalis in soybean, (ii) to determine the effect of glyphosate application timing on weed demographic parameters and soybean yield losses, and (iii) to evaluate the effect of weed density on soybean yield loss. A field experiment was conducted in two locations, in a completely randomized design with three replicates. Treatments included a control without glyphosate and glyphosate applied at soybean stages V4 or R1. The demographic stages (initial seedbank, seedlings, and adult plants) and parameters (establishment, survival, and fecundity) were estimated. Reproductive organs were evaluated in each cohort, including raceme per plant, spikelets per raceme, and seeds per spikelet. Weed and crop biomass and yield crop were assessed at harvest. Three cohorts were identified, the first of which emerged in November and contributed 93% of the total seedlings and 71% of the total adults. Glyphosate applied at V4 reduced the survival rate of the first cohort, as well as the total shoot biomass and the fecundity rate, increasing the biomass and crop grain yield. Both application timings affected tillers per plant, racemes per tiller, and fertile spikelets per raceme. Glyphosate at R1 did not effectively reduce weed competition, but reduced seed production as application at V4. Yield losses estimated with the model of the rectangular hyperbola according to weed density showed a yield loss at low densities (I) of 18%, and a maximum yield loss (A) of 82%. To avoid yield losses, herbicide applications targeting the first cohort are more effective than later applications targeting subsequent cohorts. However, at both times glyphosate applications reduced the number of seeds entering the seedbank, and therefore the population growth rate.