Sainfoin is known to vary in content of condensed tannins (CT) among accessions and harvests. In the present study, the hypothesis was tested that this results in significant differences in ruminal degradation kinetics and, with that, suitability to be used for certain purposes like increasing ruminal by pass of protein. The in sacco ruminal degradation kinetics of dry matter (DM) and crude protein (CP) of five Swiss sainfoin accessions comprising of three landraces (Moiry, Sarzens and Premier, all single flowering) and two commercial cultivars (Visnovsky and Perly, both multiple flowering) were determined with three ruminal fistulated cows. In doing so, the first and the second harvest (after 42 d of regrowth) in the second year of establishment of the plots were compared. Cultivation was done in triplicate on each of three different sites in Switzerland. Data was subjected to analysis of variance including all main effects and interactions. Finally, compositional data on CT, CP and fibre was correlated with effective ruminal degradability (ED) of DM and CP to assess how well these could be predicted. Accession hardly influenced either the degradation kinetics and effective degradability of DM and CP despite the varying chemical composition of accessions. Exceptions were the degradable but insoluble fraction, B, in DM (P=0.050) and CP (P=0.041) and the potential degradable fraction, D, in CP (P=0.017) which were influenced by accession. Conversely, harvest influenced (P<0.05) degradation kinetics and effective degradability of DM the most but not the kinetics and effective degradability of CP. The ranking of accessions was not completely consistent in the different degradation variables. However, according to the results of the multiple regression analysis, ED of DM (P=0.001) and CP (P≤0.013) declined with increasing CT content. These relationships were roughly linear. Accordingly, ED was often higher in samples from first compared to the second harvest. Most striking was, however, that only some accessions exhibited consistent and marked changes in both DM and CP degradation kinetics from the first to the second harvest whereas others did not. It appears that the first group particularly included the commercial accessions which were multiple flowering cultivars and the second group comprised the single flowering landraces. This finding could be used to select accessions with predictably stable ruminal degradability.