This study presents an integrated approach for the design of a Bicycle Sharing System (BSS) by jointly considering location decisions and capacity allocation. An important distinction of this approach is the definition of service levels, measured by the amount of unsatisfied demand both for bicycle pick-ups and returns. The method combines a set-covering model to assign location demands to stations with a queuing model to measure the related service levels. The key quality of this approach is its capacity in addressing the issues related to uncertainties in bicycle pick-up and return demand in BSS network design decisions. Results of the implementation of a BSS design for Istanbul Technical Universitya;;s Ayazaga Campus show that our approach provides a balanced BSS network by equalizing the mean demand and return rates, which will decrease the need for relocation efforts once the system is put to use. (C) 2018 Elsevier Ltd. All rights reserved.