Prognostic models for multiple myeloma have been fraught with tremendous heterogeneity in outcome among subgroups. In the context of Total Therapy 2, a tandem transplant trial for newly diagnosed myeloma, comprehensive information was available in 220 patients on standard prognostic factors (SPF), magnetic resonance imaging (MRI)-defined focal lesions, cytogenetic abnormalities (CA), fluorescence-in-situ-hybridisation (FISH)-derived amplification of chromosome 1q21 (amp1q21) and deletion of 13q14, as well as gene expression profiling (GEP). Five multivariate analysis-based survival models were derived, utilising SPF only (model 1), with progressive addition of CA (model 2), MRI (model 3), FISH (model 4) and GEP (model 5). The R(2) value, a measure of accounting for clinical outcome variability, increased progressively from 18% in model 1 to 38% in model 5. The hazard ratio for overall survival was highest for GEP (3.07, P < 0.001) followed by amp1q21 (1.71, P = 0.05). According to the presence of none (49%), one (35%) or both of these two risk features (16%), 3-year survival decreased progressively from 92% to 78% to 43% (P < 0.0001). Thus, the dominance over other prognostic parameters of molecular genetics justifies the generation of quantitative reverse transcription polymerase chain reaction methodology ('MM genetic kit') for the optimal risk stratification of patients participating in therapeutic trials.