BACKGROUND: There is a need for early identification of children with immunoglobulin A nephropathy (IgAN) at risk of progression of kidney disease. METHODS: Data on 261 young patients [age <23 years; mean follow-up of 4.9 (range 2.5-8.1) years] enrolled in VALIGA, a study designed to validate the Oxford Classification of IgAN, were assessed. Renal biopsies were scored for the presence of mesangial hypercellularity (M1), endocapillary hypercellularity (E1), segmental glomerulosclerosis (S1), tubular atrophy/interstitial fibrosis (T1-2) (MEST score) and crescents (C1). Progression was assessed as end stage renal disease and/or a 50 % loss of estimated glomerular filtration rate (eGFR) (combined endpoint) as well as the rate of renal function decline (slope of eGFR). Cox regression and tree classification binary models were used and compared. RESULTS: In this cohort of 261 subjects aged <23 years, Cox analysis validated the MEST M, S and T scores for predicting survival to the combined endpoint but failed to prove that these scores had predictive value in the sub-group of 174 children aged <18 years. The regression tree classification indicated that patients with M1 were at risk of developing higher time-averaged proteinuria (p < 0.0001) and the combined endpoint (p < 0.001). An initial proteinuria of ≥0.4 g/day/1.73 m(2) and an eGFR of <90 ml/min/1.73 m(2) were determined to be risk factors in subjects with M0. Children aged <16 years with M0 and well-preserved eGFR (>90 ml/min/1.73 m(2)) at presentation had a significantly high probability of proteinuria remission during follow-up and a higher remission rate following treatment with corticosteroid and/or immunosuppressive therapy. CONCLUSION: This new statistical approach has identified clinical and histological risk factors associated with outcome in children and young adults with IgAN.