Although a considerable body of scientific data is now available on neuroprotection in acute ischaemic stroke, this field is not yet established in clinical practice. At its third meeting, the European Ad Hoc Consensus Group considered the potential for neuroprotection in acute stroke and the practical problems attendant on the existence of a very limited therapeutic window before irreversible brain damage occurs, and came to the following conclusions. Neuroprotectants in Clinical Development: Convincing clinical evidence for an efficacious neuroprotective treatment in acute stroke is still required. Caution should be exercised in interpreting and extrapolating experimental results to stroke patients, who are a very heterogeneous group. The limitations of the time windows and the outcome measures chosen in trials of acute stroke therapy have an important influence on the results. The overall distribution of functional outcomes provides more statistical information than the proportion above a threshold outcome value. Neurological outcome should also be assessed. Neuroprotectants should not be tested clinically in phase II or phase III trials in a time window that exceeds those determined in experimental studies. The harmful effects of a drug in humans may override its neuroprotective potential determined in animals. Agents that act at several different levels in the ischaemic cascade may be more effective than those with a single mechanism of action. Current In-Hospital Management of Acute Stroke: The four major physiological variables that must be monitored and managed are blood pressure, arterial blood gas levels, body temperature, and glycaemia. The effects of controlling these physiological variables have not been studied in prospective trials, though they may all contribute to the outcome of acute ischaemic stroke and affect the duration of the therapeutic window. Optimal physiological parameters are inherently neuroprotective. Trials of new agents for the treatment of acute stroke should aim to maintain these physiological variables as close to normal as possible, and certainly within strictly defined limits. The Place of Neuroprotectants in Acute Stroke Management: Stroke patients are a very heterogeneous group with respect to stroke mechanisms and severity, general condition, age and co-morbidities. At the present time, the only firm guideline than can be proposed for patient selection is the need for early admission to enable neuroprotectant and/or thrombolytic treatment to be started as soon as possible within the therapeutic window. The severity of potential side-effects will largely determine who should assess a patient with suspected stroke and initiate treatment. There is little information on which to base the duration of neuroprotectant therapy, and more experimental data are needed. Even if prehospital treatment proves to be feasible, it should not replace comprehensive stroke management in a specialist hospital unit. Clinical trials of neuroprotectants should only be performed in stroke units. The combined approach of restoring blood flow and providing neuroprotection may be the most productive in human stroke, but current clinical trial design will have to change in order to test combination therapy. Important side-effects are those that interfere with any possible benefit or increase mortality. Pharmaco-Economic Aspects of Neuroprotectants: The early increase in hospital costs associated with neuroprotectant therapy may be balanced by the shorter length of hospital stay and lesser degree of disability of the surviving patients. The overall direct financial cost is highly dependent on the number of patients eligible for neuroprotectant therapy, which is itself dependent on the length of the therapeutic window and the severity of potential side-effects. A treatment that achieves a good functional outcome is the most cost-effective approach.