Membranous glomerulonephritis (MGN) is said to be caused by circulating autoantibodies against antigen(s) located to the epithelial side of the glomerular capillaries. Membranous glomerulonephritis with severe renal damage can be produced experimentally by injections of heterologous antibodies, but this model is obviously of questionable relevance. It can also be produced by immunizing the animal with antigen mixed with Freund's adjuvant, but this model does not prove that the damage is exerted by antibodies because Freund's adjuvant is nephrotoxic and is itself able to produce MGN. It has not yet been demonstrated experimentally that autologous antibodies alone can produce more than trace or transient proteinuria; and kidney biopsies in unselected humans have shown that MGN is a frequent finding in individuals with normal urine and normal renal function, indicating that a subepithelial formation of immune complexes is also harmless to humans. Severe glomerulonephritis without immune complex formation can easily be produced experimentally with many nephrotoxic chemicals. It is therefore suggested that the primary event in MGN, and probably other subgroups, is a toxic or allergic tubulointerstitial reaction to chemicals or drugs and that the formation of glomerular immune complexes is a later and secondary phenomenon. In agreement, renal function and the course in MGN and other glomerulonephritides are strongly correlated with the degree of tubulointerstitial damage, but totally unrelated to the degree of glomerular damage. The hypothesis explains why a large number of patients with endstage renal failure owing to glomerulonephritis have been heavily exposed to environmental pollutants.