OBJECTIVE--To examine factors determining defibrillation success and outcome in patients with ventricular fibrillation. DESIGN--Observational prospective study of age, sex, transthoracic impedance, site of cardiac arrest, ventricular fibrillation duration and amplitude, primary or secondary ventricular fibrillation, aetiology, number of shocks to correct ventricular fibrillation, and drug treatment. SETTING--A teaching hospital and a mobile coronary care unit with a physician. PATIENTS--70 consecutive patients (50 male, 20 female) mean age 66.5 years. INTERVENTIONS--Before the first countershock was administered transthoracic impedance using a 30 kHz low amplitude AC current passed through 8 cm/12 cm self-adhesive defibrillator electrode pads applied in the anteroapical position was measured. The first two shocks were 200 J delivered energy (low energy) and further shocks of 360 J (high energy) were given if required. MAIN OUTCOME MEASURES--Countershock success and outcome from ventricular fibrillation. RESULTS AND CONCLUSIONS--First shock success was significantly greater in inhospital arrests (37/53) than in out-of-hospital arrests (5/17) and in those receiving antiarrhythmic treatment (13/15, 86.7%) v (27/51, 52.9%). Transthoracic impedance was similar in those who were successfully defibrillated with one or two 200 J shocks (98.7 (26) omega) and those who required one or more 360 J shocks (91.4 (23) omega). Success rates with two 200 J shocks were similar in those patients with "high" transthoracic impedance (that is, greater than 115 omega) and those with transthoracic impedance (less than or equal to 115 omega) (8/12 (67%) v 44/58 (76%]. Fine ventricular fibrillation was significantly more common in the patients with a transthoracic impedance of greater than 95 omega (41% (13/32] than in those with a transthoracic impedance less than or equal to 95 omega (13% (5/38]. Death during arrest was significantly more common in patients who needed high energy shocks (14/18 (78%] than in those who needed low energy shocks (16/52 (31%]. Multiple regression analysis identified ventricular fibrillation with an amplitude of greater than or equal to 0.5 mV, age less than or equal to 70 years, and arrests that needed less than or equal to two shocks for defibrillation, in rank order as independent predictors of survival to discharge.