The application of a new statistical method ('breakpoint' test) to the study of the progression of chronic renal failure is described. This test establishes whether the best fit of a series of GFR measurements is linear or broken. Such an approach avoids the analytical constraint of the time of intervention assumed by other methods. Re-analysis by this test of previous studies of low-protein diet suggests that in some cases the effect of the dietary regimen has been overemphasized.