Abstract Browsing ruminants have been shown to tolerate a certain amount of tannins in their natural diet, and preference trials with captive roe deer ( Capreolus capreolus) have suggested an active selection for a low dose of hydrolysable tannins. In this study, we investigated the preference patterns for tannic acid, a source of hydrolysable tannins, and quebracho, a source of condensed tannins, in a series of preference trials with captive roe deer over time, using a pelleted feed that differed only in the respective tannin content. Additionally, two groups of four hand-raised roe deer fawns were fed either a control or a 3% tannic-acid containing diet and physiological parameters were compared after 7.5 months. There were large differences in preference patterns between the individual roe deer groups; quebracho was mostly avoided, whereas tannic acid was actively included in the diet in differing, low proportions. However, one group consistently preferred the quebracho diet over both the control or the tannic acid diet. For the tannic acid, the preference pattern often revealed an initial period of high preference, followed by a stable period of a moderate preference. The fawns on the tannic acid diet had a lower pellet intake and a higher relative mass gain than the fawns on the control diet; differences in salivary tannin-binding capacity and in blood antioxidant status were below significance. These results are the first indications of potential benefits of a low-dose tannin diet, which need further confirmation. The results of the preference trials demonstrate that the time pattern of tannin intake is not constant, and pose the question about the validity of short-term preference trials in general.