The outcome of out-of-hospital cardiac arrest (CA) following cardiopulmonary resuscitation (CPR) initiated by dispatcher-provided telephone instructions (T-CPR) in the area of Gothenburg, Sweden was studied. During a period of 27 months, 475 cases categorized by the dispatchers at the Emergency Co-ordination and Dispatch Centre as being suspected CA were offered T-CPR and were included in one of the following groups: (1) T-CPR completed (caller without previous CPR training); (2) T-CPR completed (caller with previous CPR training); (3) T-CPR started, but not completed; (4) T-CPR declined by caller due to previous CPR training; (5) T-CPR declined by caller due to other reasons; or, (6) T-CPR not offered. Of the patients, 473 could be followed up and of them 427 fulfilled the criteria for CA on ambulance arrival. Among the latter cases, 10% were hospitalized alive, 4% could be discharged from hospital, and the distribution among groups was: (1) 7%; (2) 18%; (3) 5%; (4) 11%; (5) 3%; and (6) 1%. The study concludes that although more attention should be paid to the detection of CA patients by the dispatchers, when the dispatchers suspected CA, their accuracy was high. Half of the witnesses accepted the offer of T-CPR and one-third completed T-CPR. More efforts and research are needed, however, to increase the percentages of callers completing CPR. The impact of T-CPR on survival might be limited. Indeed, the comparison of 'resuscitable' patients in whom T-CPR played an important role in supporting bystanders (i.e. groups 1 and 2) with 'resuscitable' patients in whom T-CPR was not performed (i.e. groups 3, 5 and 6) suggests an increase in survival from 6% (groups 3, 5 and 6) to 9% (groups 1 and 2).