Using an in situ approach, we have evaluated the phosphorus inputs from zooplankton in a high-mountain oligotrophic lake. Values of the specific gross release rate (SGRR) fluctuated between 02 and 2.9 ug P mg~' dry weight h~', and were higher when the nauplii of Mixodiaptomus ladniatus dominated the zooplankton community. The rate of P recycling by the zooplankton was high, reach- ing 1.6 ug P H day 1, and showed a highly consistent seasonal pattern from one year to the next, with maxima in midsummer. Zooplankton size accounted for as much as 85% of the variance obtained in the measurements of the specific rate of P release, while other factors, such as the quality or quantity of food, did not significantly influence the SGRR changes. Among the models tested, only the one pro- posed by Peters (InL Ver. Theor. Angew. LimnoL Verh., 19,273-279,1975) was useful for predictions in this system.The stoichiometric model of Hessen and Andenen (Arch. HydrobioL, 35,111-120,1992), applied in this oligotrophic system, adequately predicted the phyto- and zooplankton dynamics, whereas the values of P release estimated using this model were far higher than the excretion rates obtained experimentally. These differences were related to the type of egestion (formation of faecal pellets) of metazooplankton and to the relative importance of the food resistance to digestion. We believe that in communities where copepods constitute a substantial percentage of the zooplankton, an evaluation of the P release which is readily available (soluble reactive phosphate (SRP), total dis- solved phosphate (TDP)) to algae and bacteria would not fit the predictions of general models of mass balance; under these circumstances, assimilation efficiency proves to be the key parameter for pre- dicting the readily available P (excreted P).