We investigate the molecular first hyperpolarizability (beta) for donor/acceptor (D/A) substituted pi-conjugated organic molecules with different D/A groups and pi-systems (including up to eight pi-bonds). The results of the MP2 and density functional theory (PBE0, BMK, M05, M05-2X) calculations are compared to those obtained from experimental data. The goal of this study is to select a protocol with optimal quality/cost ratio to be used for systematic prediction of molecular nonlinear optical (NLO) properties. This goal is closely related to the way in which theoretical beta s are compared to experimental ones and to the accuracy and consistency of experimental hyperpolarizabilities used to test theoretical predictions. We found that two DFT functionals with the larger fraction of Hartree-Fock exchange (BMK and especially M05-2X) provide the best agreement to the experiment, comparable to that of the MP2 method. Due to high computational cost of MP2 method, we recommend to use the M05-2X functional as a tool for systematic prediction of molecular hyperpolarizabilities.