A type of trampling process that is caused by picking-up activities during escalator transfers was studied in this paper. A five-stage trampling model for individual pedestrians was proposed, and the social force model was modified considering the transfer features. Several scenarios were simulated to study the impacts of 4 factors, namely, pedestrian traffic, escalator velocity, picking-up duration and pedestrian velocity, on trampling probability. The results show that pedestrian traffic strongly affects the trampling probability, with a positive correlation throughout all scenarios; the picking-up duration affects the trampling probability, with a negative correlation throughout all scenarios; lower pedestrian velocities can result in higher trampling probabilities if the picking-up duration is short; and the escalator velocity may also affect the trampling probability, but there are no general rules for all scenarios. Thus, the impacts of these 4 factors can be queued in descending order as follows: pedestrian traffic > picking-up duration > pedestrian velocity > escalator velocity. Countermeasures can be employed according to the results to reduce trampling risks.