This paper presents an updated phase-shift analysis (PSA) of the low-energy (pion laboratory kinetic energy T ≤ 100 MeV) pion-nucleon (πN ) data; this solution will be referred to as 'ZRH19'. In this research programme, the modelling of the s-and p-wave K-matrix elements is achieved by following either of two methods: a) suitable low-energy parameterisations or b) the analytical expressions obtained from a hadronic model ('ETH model') based on meson-exchange t-channel Feynman graphs, as well as on s-and u-channel graphs with N and ∆(1232) intermediate states. Analytically included in the former case are the important direct (s-channel) contributions from nearby baryon resonances in the partial waves P 33 and P 11 . As it does not impose any theoretical constraints onto the data, the analysis with the Kmatrix parameterisations is suitable for the reliable identification of the outliers in the database (DB) and for the preparation of consistent input for further analysis. After the removal of the outliers, the two elastic-scattering (ES) DBs are jointly submitted to an analysis with the ETH model; the theoretical constraints of crossing symmetry and isospin invariance are imposed at this stage. The optimal values of the model parameters and the corresponding Hessian matrices are obtained from these fits, and yield Monte-Carlo predictions for the low-energy constants of the πN system, for the πN phase shifts, and for the usual low-energy πN observables. The combined π + p and π − p charge-exchange (CX) DBs are also analysed following the same procedure. During the last two decades, our PSAs make use of the Arndt-Roper formula, enabling the controlled (i.e., regulated by the reported or assigned normalisation uncertainty of each data set) rescaling of the data sets.The essential difference to our past PSAs relates to the inclusion in the DB of the two π − p s-wave ES lengths, extracted from measurements of the strong-interaction shift of the ground state in pionic hydrogen. Although the addition of these highly accurate measurements to the DB does not give rise to dramatic differences in comparison to our recent PSAs, this modification has one important consequence: the long-standing discrepancy between the value obtained (via an extrapolation to T = 0 of the π − p s-wave scattering amplitude of the ETH model) from the analysis of the ES DBs and the results extracted from pionic hydrogen is resolved. In spite of this positive development, the other discrepancies, established in our earlier PSAs of the low-energy πN data, persist; they comprise: differences in the results of the fits to the two combined DBs -i.e., to the π ± p ES DBs, and to the combined π + p and π − p CX DBs; a general difficulty (revealed after the study of the resulting χ 2 values of the various fits to the data) in the description of the π − p CX DB when imposing on the data the theoretical constraints of crossing symmetry and isospin invariance; the presence of bias in the scale factors in case of the fit to the combined π + p and π − p CX DBs; ...