A simulation model that represents a multichannel situation with a time-dependent input flow distribution is presented. The objective of the study of the system is to assess its performance under different configurations of the service stations needed to handle the peaks. A methodology is developed based on the execution of a simulation model in two steps. First, to look for the satisficing alternatives, a set of constant flow simulations is run. Second, to obtain the optimum alternative, the model is run with real flows. The presentation is based on standard queueing theory terminology and includes an example of the design of a paytoll booth in a highway in which the incoming traffic is highly seasonal. A methodology is presented to design systems that suffer short periods of very high peaks based on both queueing theory and simulation modelling. The proposed methodology follows a five-step procedure that includes obtaining the input flow and the service time distributions, a simulation model that is run both with constant and real flows and a double-graph summary tool. This double use of simulation has been proven very successful since the time spent in running the model was very small compared to the time spent in collecting the data and developing the model. Once the model was validated, it was used for a thorough experimentation period that led to a strong-based decision. A real-life example of a paytoll system in a highway is presented to validate the methodology. The limits on the number of booths are set quickly and the simulation runs help to make the final decision. Therefore, the combination of experimentation via simulation models with quantitative queueing analysis has been proven as an interesting tool to treat situations with peaks that incorporates both management input and intensive collection of data.