The CMS Collaboration has developed new "tunes" for the underlying event (UE) modeling in the PYTHIA8, PYTHIA6, and HERWIG++ Monte Carlo event generators. These tunes are constructed using different parton distribution functions (PDFs) and are based on fits to CMS UE proton-proton (pp) data at 7 TeV and CDF UE proton-antiproton (p $ \overline{p} $ ) data at lower energies. The goal is to improve predictions for proton-proton collisions at 13 TeV. The study also investigates whether the parameters obtained from UE fits are consistent with those from double-parton scattering (DPS) fits. Comparisons are made with "minimum bias" (MB) events, multijet, and Drell–Yan observables at 7 and 8 TeV, as well as predictions for MB and UE observables at 13 TeV. The underlying event consists of particles from the hadronization of beam-beam remnants (BBR), multiple-parton interactions (MPI), and their associated initial- and final-state radiation (ISR and FSR). The MPI are softer than the hard scattering processes and are regulated by a cutoff parameter $ p_{T0} $, which depends on the center-of-mass energy $ \sqrt{s} $. The parameters of the PYTHIA8, PYTHIA6, and HERWIG++ event generators are tuned to fit the UE data, with the PYTHIA8 tunes using CTEQ6L1 and HERAPDF1.5LO PDFs, and the PYTHIA6 tunes using CTEQ6L1 and HERAPDF1.5LO PDFs. The HERWIG++ tune uses a matter-overlap function related to the Fourier transform of the electromagnetic form factor. The study shows that the PYTHIA8 UE tunes, while providing improved predictions for UE data, do not describe well the data at $ \sqrt{s} = 300 $ GeV, possibly due to the modeling of the proton-proton overlap function. The PYTHIA6 and HERWIG++ tunes, on the other hand, provide better descriptions of the data. The CMS DPS tunes are constructed by fitting DPS-sensitive observables directly and calculating $ \sigma_{eff} $ from the model. These tunes are compared to other observables, including Drell–Yan and MB observables, and are shown to provide better descriptions of the data. The CMS UE tunes are validated by comparing them to UE data not used in the fits and by examining how well Drell–Yan and MB observables can be predicted. The study also shows that the CMS UE tunes can be interfaced to higher-order ME generators without additional tuning of the MPI parameters. The results indicate that the CMS UE tunes provide a good description of the UE data and that the DPS tunes areThe CMS Collaboration has developed new "tunes" for the underlying event (UE) modeling in the PYTHIA8, PYTHIA6, and HERWIG++ Monte Carlo event generators. These tunes are constructed using different parton distribution functions (PDFs) and are based on fits to CMS UE proton-proton (pp) data at 7 TeV and CDF UE proton-antiproton (p $ \overline{p} $ ) data at lower energies. The goal is to improve predictions for proton-proton collisions at 13 TeV. The study also investigates whether the parameters obtained from UE fits are consistent with those from double-parton scattering (DPS) fits. Comparisons are made with "minimum bias" (MB) events, multijet, and Drell–Yan observables at 7 and 8 TeV, as well as predictions for MB and UE observables at 13 TeV. The underlying event consists of particles from the hadronization of beam-beam remnants (BBR), multiple-parton interactions (MPI), and their associated initial- and final-state radiation (ISR and FSR). The MPI are softer than the hard scattering processes and are regulated by a cutoff parameter $ p_{T0} $, which depends on the center-of-mass energy $ \sqrt{s} $. The parameters of the PYTHIA8, PYTHIA6, and HERWIG++ event generators are tuned to fit the UE data, with the PYTHIA8 tunes using CTEQ6L1 and HERAPDF1.5LO PDFs, and the PYTHIA6 tunes using CTEQ6L1 and HERAPDF1.5LO PDFs. The HERWIG++ tune uses a matter-overlap function related to the Fourier transform of the electromagnetic form factor. The study shows that the PYTHIA8 UE tunes, while providing improved predictions for UE data, do not describe well the data at $ \sqrt{s} = 300 $ GeV, possibly due to the modeling of the proton-proton overlap function. The PYTHIA6 and HERWIG++ tunes, on the other hand, provide better descriptions of the data. The CMS DPS tunes are constructed by fitting DPS-sensitive observables directly and calculating $ \sigma_{eff} $ from the model. These tunes are compared to other observables, including Drell–Yan and MB observables, and are shown to provide better descriptions of the data. The CMS UE tunes are validated by comparing them to UE data not used in the fits and by examining how well Drell–Yan and MB observables can be predicted. The study also shows that the CMS UE tunes can be interfaced to higher-order ME generators without additional tuning of the MPI parameters. The results indicate that the CMS UE tunes provide a good description of the UE data and that the DPS tunes are