In attempting to increase the reliability of clover contribution in clover/ryegrass systems it is important to understand the roles of (1) specific traits of the clover genotype (2) climate and (3) their interactions in determining clover behaviour in swards. Overwintering and spring growth of white clover (cultivars AberHerald and Huia) grown in binary mixtures with perennial ryegrass were measured at 12 European sites ranging in latitude from Reykjavik, Iceland (64o30 ' N) to Pordenone, Italy (46°30 ' N). In the overwintering period, tiller density of the grass was assessed and detailed morphological and chemical measurements were made on the clover at each sampling time. During the growing season, the clover contribution to total available biomass was recorded. Detailed climatic data were available at all sites. The annual growth cycle of swards was divided into four functional periods (spring, summer, autumn and winter). Within each functional period community responses were modelled. The models incorporated independent biotic variables characterizing each community within each site at the start of the period and independent variables characterizing the climate at each site during the period. The models were linked dynamically by taking, as response variable(s) for a functional period, the independent biotic variable(s) of the succeeding period. In general, the modelling strategy was successful in producing a series of biologically meaningful linked models. Essential prerequisites for this were (a) the establishment of a well-devised common protocol prior to the experiment and (b) the extensive gradients of climatic and other variables obtained by using numerous sites. AberHerald generally performed as well as, or better than, Huia throughout the annual cycle across the range of climatic conditions encountered, and especially under low temperature conditions in winter and autumn. Clover leaf area index appeared to be a key variable in determining clover performance over winter and through the following growing season. Grass tiller density had a strong negative effect on clover content in spring but only at low temperatures. This emphasizes the importance of a high clover leaf area index in autumn as the main biotic factor related to spring clover content in milder conditions. The importance of climatic variables in the models is their use in explaining the reliability of the contribution of clover in clover/ryegrass systems. Temperature was the primary climatic determinant of clover response in all periods, having a direct effect on clover content and leaf area index or mediating the effect of the associate species. Radiation strongly influenced clover dynamics during winter and spring but not in the other periods, possibly because it was confounded with the effect of higher temperature. Precipitation was positively related to clover growth during spring and autumn and was related to tiller density in a complex manner during autumn and winter.