Seeds of Cenchrus ciliaris L. were produced under different hydro–photo–thermal environments with and without fertilizer. Dormancy loss of spikelets and extracted caryopses was tested during dry after-ripening at 40°C and 43% equilibrium relative humidity. Caryopses had higher initial germination and lost their dormancy faster than spikelets. Dormancy of both caryopses and spikelets generally decreased with an increase of maturation temperature and fertility, whereas dormancy increased if water stress was imposed during maturation. The latter effect was smaller when the mother plants were exposed to water stress after caryopses were fully formed than when water stress cycles were applied throughout maturation. Daylength extension (to 14 or 18 h d-1) by artificial light increased dormancy of both caryopses and spikelets. The effect of long days declined when plants were exposed to natural daylight for more than 10 h d-1. The after-ripening curves were consistent with the hypothesis that dormancy periods of individual seeds are normally distributed within each seed lot. Rates of loss of dormancy were quantified by the slopes of these curves. In a given experiment, these rates were identical for caryopses but not always for spikelets that matured in diverse environments. Even for caryopses, however, the slopes varied between experiments. Therefore, the results do not support the hypothesis that a dormancy model can be applied universally to all seed lots of Cenchrus ciliaris. Methods of predicting the period of after-ripening required to achieve desired levels of dormancy for reseeding degraded rangelands are discussed.