In a prospective controlled trial on 46 patients undergoing lumbar discectomy, three classes of variables (medical data including MRI-identified morphological abnormalities, general psychological factors and psychosocial aspects of work) were analyzed with regard to their predictive value for the outcome of lumbar disc surgery at 2 year follow-up. Multiple regression analyses were used to identify the best predictor variables of four different outcome measures (i.e. pain relief, reduction of disability in daily activities, return to work and surgical outcome). MRI-identified nerve root compromise and social support from the spouse were independent predictors of pain relief 2 years after surgery (R2 = 0.40, P < 0.01). Return to work 2 years after surgery was best predicted by depression and occupational mental stress (R2 = 0.36, P < 0.001). MRI-identified extent of herniation and depression were significant predictors of a good surgical outcome after lumbar discectomy (R2 = 0.61, P < 0.001). This study has demonstrated that the outcome of discectomy is critically dependent on which outcome variables are selected and that different sorts of predictor variables have a distinct influence on the various outcome variables. Obvious morphological alterations (i.e. disc extrusions, nerve root compromise) proved to be significant predictors of postoperative pain relief and improvement of disability in daily activities justifying a surgical treatment approach in these cases. The most important finding of this study was that return to work was not influenced by any clinical findings or MR-identified morphological alterations, but solely by psychological factors (i.e. depression) and psychological aspects of work (i.e. occupational mental stress).