We describe a novel approach to investigate and evaluate combined effect of a large number of clinical and pharmacogenetic factors on treatment outcome. We have used this approach to investigate predictors of methotrexate (MTX)-induced adverse events (AEs) leading to treatment discontinuation in rheumatoid arthritis (RA) patients. In total, 333 RA patients were genotyped for 34 polymorphisms in MTX transporters, folate and adenosine pathways. The effect of clinical and pharmacogenetic factors was assessed with penalized regression in the cause-specific Cox proportional hazards model. The predictive capacity was evaluated with the area under time-dependent receiver operating characteristic curve where cross-validation was applied. SLC19A1, ABCG2, ADORA3 and TYMS were associated with discontinuation because of AEs in clinical-pharmacogenetic model. Cross-validation showed that both clinical-pharmacogenetic model and nongenetic model had worthless predictive ability for MTX discontinuation because of AEs. These models could be further improved, either with additional polymorphisms or with epigenetic predictors.