Background. The existence of a national renal biopsy register and a national terminal uraemia status register in Denmark provides an opportunity to study the prognosis of glomerulonephritis (GN), and factors influencing prognosis. Methods. Multivariate analysis of 2380 renal biopsies with GN performed between 1985 and 1997 was done to determine the influence of clinical and histological factors on prognosis. Results. The incidence of GN (39/mio/year) and individual diagnoses did not change during the period. After 10 years, 32% were dead, 13% terminally uraemic, 5% uraemic and 50% well. Older age increased mortality, but not the incidence of renal failure after the first year. Male sex increased both mortality and incidence of renal failure (34 vs 24% at 10 years, P < 0.001). The diagnoses could be divided into three prognostic groups compared with the general population: a good prognostic group (minimal change GN and membranous GN), with a relative mortality of three and a combined renal and patient mortality of four, a poor prognostic group [crescentic GN, HUS/TTP, chronic GN] with relative mortalities of 8-19 and 13-33, respectively; and the remainder with mortalities of 4-7 and 6-12. The presence of multiple glomerular pathology, chronic GN, nephrosclerosis and chronic interstitial nephropathy worsened the prognosis, while the presence of immune deposits only worsened the prognosis of focal segmental glomerulopathy. Mortality was related to uraemia and co-morbidity at biopsy, and to the incidence of renal failure. Renal failure was correlated to uraemia and hypertension at biopsy but not to nephrotic syndrome or atherosclerosis. All vascular complications were increased and were positively related to hypertension and negatively correlated to the incidence of uraemia. Crescentric glomerulonephritis combined with anti-GBM disease had a worse prognosis than Wegener's granulomatosis, with microscopic polyangiitis and pauci-immune disease occupying an intermediate position. The prognosis of mesangioproliferative GN was unaffected by the presence of IgA nephropathy and systemic lupus erythematosus.