Aquaporins are channel forming proteins, which mediate a rapid transport of water due to the osmotic and hydrostatic pressure gradients. One aquaporin type (AQP4) has also been localized in the skeletal muscle, in which its expression is modulated by changes in muscle use. In this study we examined the sarcolemmal expression pattern and the synthesis of AQP4 in cultivated skeletal muscle fibres. The muscle fibres were isolated from musculus flexor digitorum brevis (FDB) of humanely killed rats. The isolation was performed by using collagenase treatment. Fibres were cultivated in a CO₂-independent medium (Dulbecco’s medium) overnight at 22 ± 1°C with or without Brefeldin-A (5 μg/ml), a drug that prevents transport from the ER to the Golgi elements. The muscle fibres were either immediately fixed with 3% paraformaldehyde or incubated without Brefeldin-A for 2 h at 20°C. The selected temperature prevents transport from the Golgi to plasmalemma (Matlin & Simons, 1983). AQP4 and the Golgi elements were detected immunohistochemically by using AQP4 antibody and an antibody against the endogenous Golgi marker GM-130. Samples were examined using a confocal microscope. The results revealed that the fraction of FDB fibres harbouring AQP4 varied between 33—47% in different myofibre preparations (n = 4). Moreover, the staining pattern in fibres was not uniform. Typically either one end or both ends of the fibre was stained. Homogenous staining extending throughout the fibre was found only in a fraction of fibres. BFA treatment resulted in a strong staining over z-disks and perinuclear regions extending always from one end of the fibre to the other. This staining pattern corresponds to the exit sites of the myofibre ER (Kaisto & Metsikkö, 2003). The exit site staining was found both in fibres which did not harbour AQP4 at the sarcolemma as well as in fibres in which sarcolemmal expression of AQP4 was present. After the 2 h incubation at 20°C in the absence of BFA, AQP4 was partially colocalized with the GM-130 that exhibited spots around the nuclei and in long rows of spots in the cytosol. In conclusion, the results revealed that only a fraction of the FDB fibres harboured AQP4 at the sarcolemma and in these fibres the staining pattern was not uniform. This is surprising since FDB consist almost entirely of fast twitch fibre types, in which the expression level of AQP4 should be high. On the other hand, the BFA treatment revealed that in some fibres there exists synthesis of AQP4 despite the fact that the protein is not expressed at the sarcolemma. Moreover, the exit site staining pattern was always uniform throughout the fibre indicating that all the nuclei are equally recruited for the expression of AQP4.