Modern and planned high-energy physics experiments promise to provide a lot of high-precision experimental data. The high precision is especially important in the context of searches of deviations from Standard Model predictions --- the New Physics. Consequently, the theoretical predictions should also have high precision, which in practice means going beyond NLO (1 loop) approximation. Fortunately, the multiloop calculational methods have evolved enough to provide this precision (with some reservations). However, already at NNLO level, the final results often have a very cumbersome form, which may complicate their practical use in experimental data processing. In my talk I will address the question of simplification of the results of multiloop calculations. As a rule, these results are expressed in terms of classical, harmonic, or Goncharov's polylogarithms, which all represent the examples of iterated path integrals ( or Chen's iterated integrals, not to be confused with functional integrals). I will talk about two aspects of this simplification. First, I will explain how the expression involving classical polylogarithms can be simplified using functional identities between polylogarithms. Second, I will discuss some issues related to the simplification of iterated path integrals in multivariate case.