[en] The dramatic increase in the precision of experiments measuring QCD observables requires a fresh look at high orders in QCD perturbation theory. This is a difficult problem to deal with: perturbation theory is asymptotic and it is not Borel summable. Moreover, no QCD observable has been computed beyond the three-loop order. Yet, it turns out that for many practical applications one can circumvent the problem by obtaining highly precise estimates of high orders in perturbation theory, using a recently developed technology based on Pade approximants. There are also some surprising bonuses, most notably a drastic reduction in the predicted dependence of observables on renormalization scale and scheme. The method will be reviewed, with examples from other fields and a brief description of some recent applications to concrete problems in analysis of experimental data