We aimed to study the response of Conyza sumatrensis to different doses of glufosinate, intrapopulation variation in sensitivity to the herbicide, and the heritability of phenotypic response, and model the evolution of resistance. Three studies were conducted in the greenhouse with two repetitions. First, we tested doses of glufosinate (0, 50, 100, 200, 400, 800 g a.i. ha −1) plus a nontreated check, with four replications. Second, we examined the range in sensitivity of 44 plants to 200 g a.i. ha −1 glufosinate. Third, we evaluated the sensitivity of the progeny of six glufosinate-treated plants to 200 g a.i. ha −1 glufosinate. Plant response was evaluated visually and the ammonium content in leaf tissues was measured. Glufosinate at 400 g a.i. ha −1 caused the highest injury to C.sumatrensis plants. Ammonia accumulation occurred in response to glufosinate treatment, regardless of dose. Ammonia accumulation was correlated strongly with the level of visible plant injury; thus, it is a good indicator of herbicide efficacy. Sensitivity to glufosinate was highly variable within the population. Plants with high ammonia concentration (high injury) after treatment with glufosinate produced progenies that also had high ammonia concentrations after herbicide treatment. The variation in ammonia accumulation among siblings was high. Simulating the exclusion of plants that accumulated more ammonia produced a population that is expected to be less sensitive to glufosinate in the next generation. The stronger the selection pressures by a simulated treatment with glufosinate, the greater the reduction in ammonia accumulation in the progeny and expected sensitivity to glufosinate.