Multiple sclerosis is a heterogeneous neurological disease with varying degrees of severity. The common hypothesis is that susceptibility to multiple sclerosis and its phenotype are caused by a combination of environmental and genetic factors. The genetic part exerts its effect through several genes, each having modest effects. We evaluated whether disease severity could be predicted by a model based on clinical data and data from a DNA chip. The DNA chip was designed containing several single nucleotide polymorphisms in 44 genes, previously described to be associated with multiple sclerosis. A total of 605 patients with multiple sclerosis were included in this analysis, using gender, onset type and age at onset as clinical covariates. We correlated 80 single nucleotide polymorphisms to the degree of disease severity using the following three outcome measures: linear Multiple Sclerosis Severity Score, dichotomous Multiple Sclerosis Severity Score (using a cut-off point of 2.5) and time to reach Expanded Disability Status Scale score 6. Sixty-nine single nucleotide polymorphisms were included in the analysis. No individual single nucleotide polymorphism showed a significant association; however, a combination of single nucleotide polymorphisms significantly improved the prediction of disease severity in addition to the clinical variables. In all three models the Interleukin 2 gene was included, confirming a previously reported modest effect on disease severity. The highest power was obtained using the dichotomized Multiple Sclerosis Severity Score as outcome. Several single nucleotide polymorphisms showed their added predictive value over the clinical data in the predictive models. These results support our hypothesis that disease severity is determined by clinical variables and genetic influences (through several genes with small effects) in concert.