Monitoring concentrations of organic pollutants in water is essential to predict effects and to initiate preventive steps. Results from the analysis of water samples provide snapshots of a situation, whereas monitoring using semipermeable membrane devices (SPMDs) provides a time-integrated picture of the concentration of pollutants in water. In this investigation, SPMDs, caged mussels and water samples were used to monitor the levels of organotin compounds in the inner Oslofjord, Norway, over a period of 12 weeks. The work-up procedure for the analysis of organotins was optimised, focusing on the clean-up procedure using gel permeation chromatography (GPC). By using several GPC columns, as much as 1 g of triolein could be employed. This reduces the background emission noise on the baseline, leading to an improvement in the detection limits. The main uptake of tributyltin (TBT) in mussels and SPMDs levelled off after 14 days. A longer uptake period was indicated for SPMDs at stations with a high water concentration of TBT (5-10 ng Sn L(-1)) compared with those with a low water concentration of TBT (approximately 1 ng Sn L(-1)). A concentration gradient was observed for water, SPMDs and mussels from the innermost station close to Oslo harbour to the station further out in the fjord, indicating that the three analysed matrices give approximately the same pollution gradient. The bioconcentration factor (BCF) for TBT in mussels was in the range 12-14 000 (wet weight) and, for SPMDs, 10-12 000 (fat). A good correlation with the TBT water concentrations was achieved within a period of 14-30 days of exposure for mussels and after 2-3 months for SPMDs. A good correlation was also found between the TBT concentration in SPMDs and mussels at the end of the experiment. SPMDs can therefore be used to predict concentrations of TBT in both water and mussels.