Introduction: Commonly libraries and information centers use LibQual to measure their quality of services. Although analysis of Libqual done with classical statistics, it is possible to analyze it through Artificial Neural Network with lower error rate. This research try to introduce an Artificial Neural Network that is able to predict s ervice quality of university library.

  Methods: In this applied cross-sectional study, all of Shiraz university of medical science students were assessed. LibQual questionnaire was the instrument of data collection and MATLAB software was being used to analyze data. In addition an algorithm was written to automatic selection of the best network architecture based on lower error rate and higher adaptation rate.

  Results: for 5 categories of input data and with running of the written algorithm, 5 ANN was created and their matching ratio is 0.77059, 0.6828, 0.81089, 0.79161 and 0.83273 respectively.

  Conclusion: By comparing the ANNs, it was found that ANN with 20 hidden layer, %80 training data, %16.667 testing data and %3.3333 validation data that be fed with fifth input data, is the most appropriate ANN in quality evaluation of university libraries.