We have previously demonstrated a 60-80% ischemic loss of somatostatinergic neurons in the dorsal dentate hilus of the rat. However, several studies have failed to demonstrate ischemic loss of GABAergic neurons in hilus, although one study reports that 96% of the somatostatinergic neurons in the dorsal hilus colocalize GABA. In order to understand this paradox, we have now estimated, using unbiased stereology, the total number of neurons immunohistochemically stained against glutamic acid decarboxylase-65 (GAD65) and GAD67 in the dorsal dentate hilus. Rats were divided into groups subjected to either sham operation (n=8) or 8 min of transient global ischemia during systemic hypotension (n=8) and allowed to survive for 7-9 days. Results from cell counts (mean +/- SD) in sham rats demonstrated that the dorsal hilus contains 9,189+/-3,957 GAD65 neurons and 6,991+/-2,784 GAD67 neurons. After ischemia, corresponding cell counts demonstrated 10,216+/-4,866 GAD65 neurons and 10,119+/-5,906 GAD67 neurons, and these results were not significantly different (P>0.05) from results in sham rats. Power analysis of the t-test informs that losses less than 80% are not significant and reflects the excessive variance in our material. For comparison, we estimated a total of 21% ischemic neuron death in the dorsal hilus on cresyl violet-stained sections from other corresponding sham (n=7) and ischemic rats (n=7). This explains why ischemic loss of hilar GABAergic neurons can only be detected by counts of the vulnerable subpopulation colocalizing somatostatin. Our investigation has demonstrated a surprisingly high variation between rats in a number of GAD-immunopositive neurons located in the dorsal dentate hilus, which is related to variations between the individual rats and neurons in their endogenous GAD expression.