The overall aim of this thesis was to try to predict individuals with Juvenile Idiopathic Arthritis (JIA) who do not respond to a drug called methotrexate (MTX). The clinical information from individuals recruited to four national JIA studies was combined for collective analysis. Genetic data across the entire genome was generated for those who had either a blood sample or saliva sample collected as part of the study. Non-response to MTX was defined using a standard measure. Clinical and genetic information was available for 2211 and 1199 cases. Almost one-third of the individuals did not respond to MTX at six months. The clinical factors that could predict non-response included absence of a protein in the blood called antinuclear antibody, lower measures of disease at start of MTX and taking MTX by mouth instead of injection. However, all the clinical factors combined had a low accuracy of prediction especially for it to be clinically useful. None of the genetic factors were able to predict nonresponse to MTX and when the genetic factors were combined with the clinical factors the accuracy of the prediction did not improve. However, starting MTX by injection route over the oral route can be recommended.