Although a disruptive effect of low-level microwave (MW) radiation on the blood-brain barrier (BBB) of small laboratory animals has been suggested by a number of authors, the relevance of these studies to larger animals has received little attention. Using 131I albumin as our tracer molecule and cannulation of the cisterna magna and of the femoral vein to permit multiple sampling of cerebrospinal fluid (CSF) and plasma, we have studied the effects of microwave radiation on the BBB in dogs. Measurement of the CSF/plasma distribution ratio (cmpCSF/cmpplasma) of 131I albumin was carried out over a 5-hour period after exposure of the dog's head for 20 minutes to various power densities of continuous wave microwave at a constant frequency of 1.0 GHz. Control animals (n = 11) were subjected to the same experiment, but received no MW exposure. No effect on the BBB was observed in two dogs each at the following power densities: 2, 4, 10, 50, and 200 mW/cm2. Eleven dogs were exposed to 30 mW/cm2, and, within this group, there were two animals in which the penetrance of the BBB by the 131I albumin was increased 4–5 times over controls, and two additional animals which showed twofold to fourfold increased penetrance. However, the CSF : plasma distribution ratios of the remaining animals receiving 30mW/cm2 did not differ statistically from controls. The results obtained in the four dogs experiencing a postmicrowave alteration of the blood-brain barrier suggest that, for the dog, a microwave ‘window’ effect may be operating very near 30 mW/cm2.