The study of restenosis after angioplasty poses serious methodological problems. The first is the definition of angiographic criteria of restenosis. These should be based on quantitative angiographic measurements in absolute values of coronary diameter rather than on the use of percentage stenosis which is an inaccurate indication of the true severity of the coronary disease. Moreover, the use of an arbitrary threshold > or = 50% stenosis at angiographic control tends to "pre-select" poor initial results of angioplasty as restenosis. Criteria based on absolute values of coronary artery diameter have enabled the demonstration of a close correlation between an excellent result of angioplasty and the degree of the restenosis 6 months later which suggests that a too good result of angioplasty may be related to increased intimal hyperplasia. This is a real dilemma for those performing angioplasty knowing that a mediocre initial result does not guarantee a good long-term result. In addition, it seems that the diameters of coronary arteries 6 months after angioplasty have a Gaussian distribution. This would imply that intimal hyperplasia is a constant phenomenon after angioplasty and that it is its degree which varies between patients with and without restenosis. Restenosis would therefore be more of a quantitative than a qualitative phenomenon. This justifies the use of continuous variables in the study of restenosis and a categorical approach would therefore be less valuable, not as powerful statistically and based on thresholds of an arbitrary nature. This could also explain the contradictory results concerning predictive factors of restenosis in the literature.(ABSTRACT TRUNCATED AT 250 WORDS)