The purpose of this study was to examine the usefulness of digital subtraction angiography in the evaluation of left ventricular function (LVF). LVF was examined in 24 patients by three methods, (a) conventional ventriculography using an intraventricular injection (CLV) of 40 ml of contrast medium, (b) small volume (10-15 ml) left ventriculogram (SVLV) and (c) intravenous injection of 30 ml contrast medium (IVLV). Images were recorded via a conventional image intensifier--TV chain on to video tape and later analysed using a nuclear medicine computer with a digital interface. There was excellent correlation of ejection fraction and wall motion abnormalities between CLV and SVLV methods (r = 0.92 and r = 0.71, respectively) and good correlation between CLV and IVLV methods (r = 0.88 and r = 0.67, respectively). However, only minimal contrast induced premature ventricular contractions were observed in either the SVLV or IVLV methods compared with approximately 60% in the CLV method. Exclusion of studies containing ectopic cardiac cycles considerably improved the correlation between CLV and SVLV (r = 0.97) and between CLV and IVLV methods (r = 0.95). In conclusion, our results indicate that conventional left ventriculography may be replaced by either intravenous or small volume intraventricular methods with little loss of wall motion definition or error in ejection fraction estimation. The less invasive nature of the SVLV and IVLV methods should increase the usefulness of left ventriculography and extend its application to the assessment of left ventricular reserve under stress and during drug intervention.