In a double blind prospective randomized study 68 patients (ASA I and II) aged 20-40 years, undergoing elective general, orthopaedic, or gynaecological surgery were selected. After the measurements of baseline haemodynamic variables, all patients were given fentanyl 1 mu gkg(-1) followed by either 0.3 mgkg(-1) ketamine (group KP), midazolam 0.03 mgkg-1 (group MP), propofol 0.4 mgkg(-1) (group PP) or, normal saline 3 ml (group SP -control). Induction of anaesthesia was done by titrated dose of propofol preceded by 2 ml of lignocaine. Using loss of response to verbal commands as end point of induction, the dose of propofol required to induce anaesthesia was significantly lower in group KP (1.2 mgkg(-1)), MP (1.4 mgkg(-1)), and PP (1.6 mgkg(-1)) compared to control group (2.7 mgkg(-1)). Fall in mean arterial pressure (MAP) from the baseline following induction was observed in all the groups being maximal (21%) in control group and minimal (4%) in group KP. Relative bradycardia was seen in all patients, but least in KP group. The group MP and PP had 13% and 11% falls in MAP respectively. We conclude that all co-induction agents reduce the requirement of propofol compared to placebo and haemodynamic effects were dose dependent. Ketamine appears to be a suitable and safe alternative to midazolam co-induction. Propofol auto-co-induction does not offer any advantage over midazolam regarding cardiovascular stability.