For the first time a complete set of the most recent direct data on primary cosmic ray spectra is used as input into calculations of muon flux at sea level in wide energy range $E_\mu=1-3\cdot10^5$ GeV. Computations have been performed with the CORSIKA/QGSJET and CORSIKA/VENUS codes. The comparison of the obtained muon intensity with the data of muon experiments shows, that measurements of primary nuclei spectra conform to sea level muon data only up to several tens of GeV and result in essential deficit of muons at higher energies. As it follows from our examination, uncertainties in muon flux measurements and in the description of nuclear cascades development are not suitable to explain this contradiction, and the only remaining factor, leading to this situation, is underestimation of primary light nuclei fluxes. We have considered systematic effects, that may distort the results of the primary cosmic ray measurements with the application of the emulsion chambers. We suggest, that re-examination of these measurements is required with the employment of different hadronic interaction models. Also, in our point of view, it is necessary to perform estimates of possible influence of the fact, that sizable fraction of events, identified as protons, actually are antiprotons. Study of these cosmic ray component begins to attract much attention, but today nothing definite is known for the energies $>40$ GeV. In any case, to realize whether the mentioned, or some other reasons are the sources of disagreement of the data on primaries with the data on muons, the indicated effects should be thoroughly analyzed.