Stroke causes excitotoxicity, inflammation, cell death and compensatory neurogenesis. Resident and infiltrating cells generate nitric oxide (NO) as a result of the activation of NO synthases (NOS)¹. Experimental cerebral ischemia leads to the up-regulation of all three NOS isoforms, although the patterns of expression are temporally and spatially distinct. Classic gene knockout studies indicate that the activation of NOS-1 is detrimental, while the NO derived from NOS-3 is beneficial. Expression of NOS-2 is induced in both resident and infiltrating cells later than either NOS-1 or NOS-3, and this NO does not contribute to early pathology. Significant increases in expression of pro-inflammatory cytokines within a few hours of ischemic onset are assumed to trigger transcriptional activation of NOS-2. However, the promoter contains a hypoxia response element such that HIF-1α can activate the gene, and some neuroprotectants block cytokines without affecting NOS-2 expression². Furthermore, transcriptional activation may not account for the very rapid appearance of NOS-2 positive cells infiltrating the infarct. While gene knockout studies suggest that NO produced by NOS-1 and NOS-2 is detrimental, and that derived from NOS-3 is beneficial, studies employing various NOS inhibitors provide conflicting results³. Collectively, NOS inhibitors significantly reduce infarct volume. Treatment before stroke onset reduces infarct volume in transient models, while early administration of NOS inhibitors after ischemic onset is effective in permanent stroke. Later treatment has a beneficial effect on infarct volume in both types of stroke model. Non-selective inhibitors do not alter infarct volume in permanent ischemia, whereas ‘‘selective’’ NOS-1 and NOS-2 inhibitors reduce lesion size regardless of experimental model. It is likely that the beneficial effects of non-selective inhibitors are limited because they inhibit NOS-3 to a similar degree. Consequently, they may aggravate ischemia by increasing platelet aggregation and white cell activity, raising blood pressure, and by restricting penumbral blood supply. Exogenously applied arginine (NOS substrate) has produced conflicting results: decreasing, having no effect, or even increasing infarct volume⁴. The same is true for the effect of arginine on cerebral blood flow which, if enhanced, could rescue salvageable tissue from the spreading ischemic core. This may be due to the ability of arginine to enhance NO from all three NOS isoforms. Conversely, experimental studies and human trials with NO donors show beneficial effects following experimental ischemia. However, unopposed high doses of NO donors might be detrimental due to profound vasodepressant effects. Following ischemic brain injury there is an increase in the rate of neurogenesis. These new cells arise in the dentate gyrus and subventricular zone, and populate subcortical and to a lesser extent cortical structures. While NO is reported to be cytostatic, or to promote terminal differentiation of neural stem cells in the uninjured brain, it appears to be anti-apoptotic in the ischemic brain. Administration of NO donors can enhance angiogenesis and neurogenesis, and significantly improve functional outcome. Mice deficient in NOS-2 do not display the predicted neurogenic response following ischemia. The mechanism is unknown but it could be that the resultant inflammatory reaction in the absence of NOS-2 derived NO proceeds unabated, resulting in an environment detrimental for neural progenitor cell differentiation and survival. While the generation of NO clearly contributes to pathology following ischemia it can also be beneficial⁵. Explanations for this apparent contradiction may be found in the cell type responsible, the amount of NO that is generated, and whether products of further NO oxidation are formed. A number of important questions have yet to be answered with respect to the roles for the NO generated by different cell types following acute injury. Seeking answers through the use of NOS inhibitors, or even knockout mice, will not be definitive. Conditional and/or cell-specific knockouts, in which NOS responses can be selectively manipulated, will be useful models for unpicking NO and neuropathology.