Short-term field experiments are often used to predict and evaluate long-term management effects. Based on a mowing experiment in two calcareous fens near Lake Neuchatel, Switzerland, we investigated whether shea-term treatment effects (i.e. during the first four years) were confirmed by longterm results (13 - 14 yr). Plots were mown in summer or in winter or left unmown. The main long-term trends in overall species composition (based on percentage cover estimates) were already observable in the first four years: mown and unmown plots diverged, whereas summer-cut and winter-cut plots remained similar. At the individual species level, however, short-term and long-term treatment effects differed considerably: many species whose abundance seemed affected by treatments during the first four years showed no response in the long term, and vice versa. These discrepancies were similar when based on cover estimates or on counts of shoots. Species responses did actually depend on the time scale considered. Short-term and long-term treatment effects on species richness were similar (i.e. a decrease in unmown plots), although only long-term effects were significant. Treatment effects on the above-ground biomass varied considerably, and short-term trends (lower biomass in unmown plots) differed from long-term trends (higher biomass in unmown plots). Our sites showed little overall change in species composition during the period investigated, and treatment effects were low compared with other similar experiments. if study sites are less stable or treatment effects more drastic, a short-term evaluation is expected to be even less reliable. Knowledge on species dynamics at a site may help to choose the adequate spatial and temporal scale of investigation, and thus increase the efficiency of management experiments.