Chemical mutagenesis is a useful tool for inducing mutations in plants. Seeds are often used as the material for chemical mutagenesis. The biological effect of a chemical mutagen on seeds is determined by absorption dose (the product of mutagen concentration and acting time, which starts after the mutagen is absorbed by the seeds). In practice, however, the concept of exposure dose (the product of mutagen concentration and treating time) is usually used instead because the time for absorbing mutagen is unknown. In this study, we conducted an experiment using ethyl methane sulphonate (EMS) to treat cauliflower seeds, in which five EMS concentrations (0%, 0.5%, 1.0%, 1.5% and 2.0%), three treating time lengths (4 h, 6 h and 8 h) and two pretreatments (non-presoaking and presoaking of seeds for 2 h) were set. We obtained a well-fitted nonlinear regression model for the relationship between seedling survival rate and the EMS treatment, and its marginal models for the two pretreatments. Based on the models, we determined the EMS absorption doses under the two different pretreatments and identified their 50% lethality dose (LD50). We found that presoaking could delay EMS absorption and therefore reduce the injury caused by EMS within a given treating time, but could hardly change the biological effect of EMS after it is absorbed. The conclusions about absorption dose and presoaking effect obtained in this study might be generally applicable to plant chemical mutagenesis in principle.