Consumer care products often contain UV filters, organic compounds which absorb ultraviolet light. These compounds may enter surface waters directly (when released from the skin during swimming and bathing) or indirectly via wastewater treatment plants (when released during showering or washed from textiles). Predicted and measured UV filter concentrations were compared in a regional mass balance study for two Swiss lakes: Lake Zurich, a typical midland lake which is also an important drinking water resource, and Hüttnersee, a small bathing lake. Both lakes are extensively used for recreational activities and considerable direct input of UV filters is thus expected. This input was estimated from the number of visitors at swimming areas around the lakes and a survey of the usage of sunscreen products among these visitors. Possible additional indirect input via wastewater treatment plants was not considered in this study. The quantitatively most important UV filters, as indicated by the survey data, ethylhexyl methoxycinnamate, octocrylene, 4-methylbenzylidene camphor, butyl methoxydibenzoylmethane, and benzophenone-3, all lipophilic compounds, were selected for analysis by gas chromatography–mass spectrometry. Concentrations of individual UV filters in water from Lake Zurich were low, ranging from <2 ng l−1 (detection limit) to 29 ng l−1, and somewhat higher at Hüttnersee, ranging from <2 to 125 ng l−1, with the highest concentrations found in summer, consistent with direct inputs to the lakes during this time. The concentrations were clearly lower than predicted from input estimates based on the surveys. This may be in part due to (i) an overestimation of these inputs (e.g. less than the 50% wash-off of UV filters assumed to occur during swimming), and (ii) some removal of these compounds from the lakes by degradation and/or sorption/sedimentation. UV filters were also detected in semipermeable membrane devices (SPMDs) deployed at Lake Zurich and Greifensee, another midland lake, at concentrations of 80–950 ng SPMD−1, confirming the presence of the compounds in surface waters and indicating a certain potential for bioaccumulation. SPMD-derived water concentrations were in the range of 1–10 ng l−1 and thus corresponded well with those determined in water directly. No UV filters were detected above blank levels in SPMDs deployed at a remote mountain lake used for background measurements.