Aims: Recent studies have suggested that patients with coronary disease suitable for angioplasty have an equally good outcome with medical therapy if clinically stable. Complex lesion morphology may predict acute events without intervention and stenosis severity influences the degree of collateralisation. This study was designed to assess the influence of these factors on clinical outcome. Methods and results: A retrospective review of patients suitable for angioplasty who were randomised to initial medical therapy as part of a multicentre study. Angiograms were reviewed for lesion characteristics, TIMI flow grade, and degree of collateralisation. Angiograms were available on 79 patients (13 female, 66 male). Mean age was 54.8 years (range 43–68) in the group crossing-over to revascularisation, and 58.4 (range 37–78) in the group who did not ( P=ns). Seventeen patients crossed-over (two to CABG, 15 to PTCA) at 5.4 months (range 0–10) after initial angiography. Disease progression had occurred in 10/17 patients (58.8%), three of whom developed a new occlusion. Collateralisation was more likely in smokers, independent of lesion severity ( P<0.05). Time to cross-over was not influenced by progression of disease. Crossing-over was not affected by age, diabetic status, cholesterol level, vessel involved, lesion severity, TIMI flow, lesion morphology, collateralisation, or the number of vessels diseased, but was more likely in females ( P<0.05). Conclusion: This group of patients generally does well with medical therapy. Whilst the numbers are relatively small, there does not appear to be any reliable prospective marker, including the presence of spontaneous collateral channels on diagnostic angiography, to indicate which patients will fail medical therapy and require revascularisation.