Over time, stand density reduction in Norway spruce (Picea abies (L.) Karst.) (a) reduces total yield degressively and increases dominat diameter propressively, (b) reduces total yield less through thinning from above than through thinning from below, (c) reduces total yield more severely on rich compared to poor sites (SI 45 vs. SI 25), (d) accelerates dominant diameter more on rich compared to poor sites (stand age always 100 y, SI 40 in (a) and (b)). Display omitted •Maximum yield in unthinned stands and only short-term growth acceleration by thinning.•Strongest size growth acceleration but total yield losses on rich sites.•Strongest total yield losses by thinning from below. Many current stand management guidelines propose low initial stand densities and strong density reductions in order to lower the costs of stand establishment, to accelerate stand growth and promote the diameter growth of selected future crop trees. The long-term effects of density reductions on growth and yield, however, are often neglected; they remain open for debate due to a lack of empirical evidence. Here we examine 22 thinning experiments in Norway spruce (Picea abies (L.) Karst.) located in Germany with 127 plots and 1209 full stand measurements to revisit the density-growth relationship. These experiments cover both short- and long-term growth reactions to thinning since their establishment in 1882. First, we show the temporarily unimodal optimum relationship between periodical increment and stand density; and how it results in a saturation curve between stand yield and stand density in the long term. We particularly highlight how the effect of stand density reductions on growth reduces across stand development. Second, we show the dependency of total yield on the thinning (kind, severity and intensity of thinning) and site quality. Over time, unthinned stands achieve the highest total yield of stem wood. Thinning causes severe growth losses, especially on rich sites and through thinning from below; e.g. on top sites a continuous density reduction by thinning from below to 50% of the maximum density reduced the total yield by 26% or 670 m3 ha−1 till the age of 100 years. Third, we demonstrate that the effect of thinning on the diameter of dominant trees is strongest on rich sites and similar when thinned from above or below. Over time, accelerating diameter growth incurs a high cost in terms of stand yield. Finally, we examine the relevance of our results to population ecology and production economy. We discuss the superior yield after thinning from above, the tradeoff between tree diameter growth acceleration and yield, and the relevance of long-term experiments and their impact on silvicultural prescriptions.