Surgery speeds up recovery for sciatica. Prolonged conservative care with surgery for those patients with persistent sciatica however, yields similar results at 1 year. To investigate whether baseline variables modify the difference in recovery rates between these treatment strategies, baseline data of 283 patients enrolled in a randomized trial, comparing early surgery with prolonged conservative care, were used to analyse effect modification of the allotted treatment strategy. For predictors shown to modify the effect of the treatment strategy, repeated measurement analyses with the Roland Disability Questionnaire and visual analogue scale pain as continuous outcomes were performed for every level of that predictor. Presumed predictive variables did not have any interaction with treatment, while "sciatica provoked by sitting" showed to be a significant effect modifier (P = 0.07). In a Cox model we estimated a hazard ratio (HR, surgery versus conservative) of 2.2 (95% CI 1.7-3.0) in favour of surgery when sciatica was provoked by sitting, while the HR was 1.3 (95% CI 0.8-2.2) when this sign was absent. The interaction effect is marginally significant (interactions are usually tested at the 10% level) but the patterns generated by the repeated measurement analyses of all primary outcomes are completely consistent with the inferred pattern from the survival analysis. Classical signs did not show any contribution as decision support tools in deciding when to operate for sciatica, whereas treatment effects of early surgery are emphasized when sciatica is provoked by sitting and negligible when this symptom is absent.