After an initial demineralisation, an intensive fluoridation is capable of inhibiting the erosive mineral loss in dentine completely, which might be related to the presence of the exposed organic dentine matrix. Aim of the present study was, therefore, to evaluate whether fluoride is also effective when the organic material is removed. The study was a cyclic de- and remineralisation model over 5 days in vitro. Samples from human coronal dentine were demineralised with citric acid (pH 2.3; 6 x 10 min per day) and intermittently stored in a remineralisation solution. Groups (n = 25 each) were defined as follows: Group 1: erosion only, no fluoridation; Group 2: erosion, alternately fluoridation with toothpaste (NaF; 0.15% F-; 3 x 5 min per day), mouthrinse (Olaflur/SnF2; 0.025 F-; 3 x 5 min per day) and gel (Olaflur/NaF; 1.25% F-; at Days 1 and 3 instead of one toothpaste application); Group 3: erosion and fluoridation as Group 2, organic material was continuously removed with collagenase (from Clostridium Histolyticum type VII, 100 U/ml) added to the remineralisation solution. Mineral content was monitored daily using longitudinal microradiogaphy. After fluoridation in the presence of collagenase, a linear increase in mineral loss (73.3 +/- 17.6 microm at Day 5) was observed, which significantly (P < or = 0.001) exceeded that of the control group (45.9 +/- 14.3 microm at Day 5). After fluoridation in the absence of collagenase, mineral loss ceased after the 2nd day (12.2 +/- 10.2 microm at Day 5) and was significantly lower compared to Groups 1 and 3 (P < or = 0.001). The results indicate that the dentine matrix is essential for the effectiveness of fluoride in dental erosion.