Tree growth is strongly influenced, among other factors, by climate. Much knowledge regarding climate growth relationships has been gained by studying tree rings. However, sufficient tree-ring data is rarely available if climate effects are required to be representative for large spatial scales, for example, in order to be used in scenario models to estimate forest development under climate change. Alternative data sources include large-scale forest inventories, although these usually provide lower temporal resolutions than tree rings. When working with temporally sparsely-resolved growth data, the question of how climate growth relationships aggregate over time becomes relevant. To overcome this trade-off between spatial representativeness and temporal resolution, this study aims at optimally using the information contained in the annual resolution of tree rings to derive recommendations regarding the choice of climate variables for modeling tree growth based on forest inventories. We evaluated for Picea abies and Fagus sylvatica, which part of the year (spring, summer, vegetation period, whole year) and whether mean or extreme climatic conditions within inventory intervals should be taken into consideration. A three-step approach was used: (1) we used response functions to quantify the effect of monthly precipitation and temperature on annual basal area increments, (2) we temporally aggregated the annual basal area increments to hypothetical intervals of five and ten years, and correlated them with climate means and extremes – from different parts of the year – within the aggregated intervals, and (3) we fitted linear mixed-effects models to simultaneously quantify the effects of the climate variables, site characteristics and the years of the hypothetical inventories. The results did not generally differ between both species. Variables based on conditions during the whole year and partly during spring performed better than variables based on conditions during summer or the vegetation period. Defining the year as the period between October of the previous year and September of the current year allows possible lag effects of previous autumn and winter conditions to be taken into consideration. Mean climatic conditions reached or exceeded the correlations of the extremes and mostly performed similar to or better than the extremes in the models. Our results indicate that these relationships are insensitive to the often arbitrarily determined years, in which inventories take place. These findings can serve as basic recommendations for the choice of climate variables when modeling climate effects on multi-year growth of P. abies and F. sylvatica in the European lowlands.