Prognostic and mechanistic schemes for the determination of plant phenological stages from environmental conditions and the estimation of net primary production (NPP) are presented. The new schemes account for different biomes and are included in a global model of carbon cycling in the terrestrial biosphere. The capability of such a model to simulate the seasonal cycle of atmospheric CO2 is explored. The model is forced by mean monthly climate variables (temperature, precipitation and light) and the mean annual CO2-concentration. It predicts the atmosphere{biosphere CO2 exchange fluxes, leaf area index (LAI), and the times of budburst and leaf abcission. The predicted variables offer means of validation against data of the observed annual cycle of atmospheric CO2- concentration and observations of LAI derived from satellite data. Estimated annual NPP of forests appears realistic, however NPP of grass dominated biomes is greatly underestimated. This seems to be related to the fact that belowground biomass is not explicitly considered in the model. The results of a simulation of the seasonal cycle of of [sic] atmospheric CO2-concentration using a three dimensional atmospheric transport model are in satisfying agreement with the observations. ; This is the authors' final draft of the paper published as Climate Research, 1996, 6(1), pp. 1-19. Contents page of the issue available at http://www.int-res.com/abstracts/cr/v06/n1/