In vivo administration of pertussis toxin is often used to study the involvement of guanine nucleotide binding proteins in signal transduction. Especially when it is administered in the brain the effect is often poor. This could be due to the fact that pertussis toxin does not reach the area of interest. To evaluate the extent to which pertussis toxin is distributed in rat brain after intraventricular injection, different techniques were used. Immunohistochemical studies with an antibody against pertussis toxin showed that immunoreactivity was limited to periventricular brain structures less than 0.5 mm from the lumen. The highest immunoreactivity was seen 16-24 h after injection. After 96 h the labeling was very weak. The proportion of guanine nucleotide binding proteins that were ADP-ribosylated by in vivo injection of pertussis toxin into the ventricles as assessed by in vitro [32P]-back-ADP-ribosylation was very low 48 h after the injection, in all regions studied. Direct injection of pertussis toxin into the brain caused a marked ADP-ribosylation localized to the region injected that was maximal at 72 h after injection. At 96 h there were also effects after control injections, indicating non-specific effects. Synaptosomal membranes and other membranes were equally affected by pertussis toxin. The results suggest that in studies regarding the effect of pertussis toxin treatment on signal transduction, the toxin must be injected very close to the brain region of interest and, furthermore, that the rats should be killed 48-72 h after injection. In case of lack of effect on the response of interest one should examine whether the ADP-ribosylation of pertussis toxin-sensitive guanine nucleotide binding proteins in the area of concern has been affected.