The fate of exogenous ganglioside GM1 labelled in the sphingosine moiety, [Sph-3H]GM1, administered as a pulse, in the isolated perfused rat liver was investigated. When a non-recirculating protocol was employed, the amount of radioactivity in the liver and perfusates was found to be dependent on the presence of BSA in the perfusion liquid and on the time elapsed after the administration of the ganglioside. When BSA was added to the perfusion liquid, less radioactivity was found in the liver and more in the perfusate at each time tested, for up to 1 h. The recovery of radioactivity in the perfusates followed a complex course which can be described by three pseudo-first-order kinetic constants. The constants, in order of decreasing velocity, are interpreted as: (a) the dilution of the labelled GM1 by the constant influx of perfusion liquid; (b) the washing off of GM1 loosely bound to the surface of liver cells; (c) the release of gangliosides from the liver. Process (b) was found to be faster in the presence of BSA, probably owing to the ability of BSA to bind gangliosides. The [Sph-3H]GM1 in the liver underwent metabolism, leading to the appearance of products of anabolic (GD1a, GD1b) and catabolic (GM2, GM3) origin; GD1a appeared before GM2 and GM3 but, at times longer than 10 min, GM2 and GM3 showed more radioactivity than GD1a. At a given time the distribution of the radioactivity in the perfusates was quite different from that of the liver. In fact, after 60 min GD1a was the only metabolite present in any amount, the other being GM3, the quantity of which was small. This indicates that the liver is able to release newly synthesized gangliosides quite specifically. When a recirculating protocol was used, there were more catabolites and less GD1a than with the non-recirculating protocol. A possible regulatory role of ganglioside re-internalization on their own metabolism in the liver is postulated.