I agree with Dr Watts that observational studies do not provide definitive answers. Unfortunately, observational studies are currently all that clinicians have in the area of HRT and the primary prevention of CHD in postmenopausal women, and while they do not enable us to draw definitive conclusions, they do enable us to make provisional inferences about the effects of therapy on a study-by-study basis. This is common practice in translating findings from epidemiological research into decisions in clinical practice. It is for this reason that I described a reduction in the incidence of CHD rather than a lower incidence of CHD in postmenopausal users of HRT. Standard terminology, such as beneficial effect, used to describe a lower incidence reflected the authors' use of these terms in the articles cited (eg, Grodstein et al1 and Sellers et al2). I do not believe that my phrase implies causality, although as an epidemiologist and clinician, I respect both viewpoints.