Geopotential height fields for nine winters (1967/68–1975/76) and summers on three levels (300, 500, 1000 mb), provided by the German Weather Service, are used to compute daily budgets of kinetic energy (Km) for single zonal wavenumbers m within a quasi-geostrophic framework. The energetics are averaged over the depth of the troposphere and the latitudinal belt 40°N ≤ ϕ ≤ 85°N. Inadequate field resolution by data, and inevitable model approximations justify regressive corrections of the energy conversion estimates, leading to upper-bound estimates of the fraction of kinetic energy variance explained by different conversions (hindcast skills). The results are (i) Persistence is most important for the evolution of Km for m ≤ 4; the skill ranges from 76% to 59% (in winter). The importance is less for m ⩾ 5, according to the skills ranging from 51% to 43% (in winter). (ii) For 1 ≤ m ≤ 4 the sum of energy conversions explains approximately 35%–45% of the variance of Km, left unexplained by persistence in winter. In this wave range the leading process is the nonlinear interaction with waves m ≤ 5, whereas conversions between Kz and Km as well as lateral boundary fluxes are of the least importance. The evolution of Km for m = 2,…,4 is unexpectedly well correlated to the conversion between available potential energies C(AzAm) which appears to be a good substitute for C(AmKm), which is proportional to the vertical sensible heatflux. (iii) For waves with 5 ≤ m ≤ 8 the sum of conversions explains 30%–50% of the variance of Km, left unexplained by persistence in winter. The baroclinic conversion C(AmKm) contributes the largest skill; however, C(AzAm) is a better indicator of baroclinic activity even in this wave range. The next largest impact on Km stems from nonlinear wave- wave interactions involving at least one other short wave with m = 6,…,12. The influence of C(KzKm) is insignificant in the wave range 5 ≤ m ≤ 8. (iv) The total skill of our model (including persistency) ranges from 65% (synoptic scales) to 85% (ultralong waves) in winter. (v) In summer, the amount of variance of Km, explained by different conversions, is generally lower than in winters, whereas the overall scale-dependent skill patterns are essentially preserved.