The structure of premixed ammonia + air flames, burning at atmospheric pressure under strain-stabilized conditions on a porous-plug burner, has been investigated using laser-diagnostic methods. Profiles of OH, NH, and NO were acquired by laser-induced fluorescence (LIF) and quantitative concentrations of OH and NO were retrieved using a concept for calibration versus absorption utilizing the LIF-signal itself, whereas NH concentrations were evaluated employing a saturated fluorescence signal. In addition, temperatures and relative oxygen concentrations were measured by rotational Coherent Anti-stokes Raman Spectroscopy (CARS). The new experimental data for flames with equivalence ratios of 0.9, 1.0, and 1.2 were used to validate and rank the performance of four contemporary detailed kinetic models. Simulations were carried out using experimental temperature profiles as well as by solving the energy equation. Two models of the same origin, developed by Mendiara and Glarborg (2009) and by Klippenstein et al. (2011), in most cases showed good agreement in terms of radical concentrations, however, the model of Mendiara and Glarborg had better prediction of temperatures and flame front positions. The model by Shmakov et al. (2010) had comparable performance concerning radical species, but significant discrepancies appeared in the prediction of flame front positions. The model of Duynslaegher et al. (2012), in addition to the flame front positions, deviated from experiments or other models in terms of NH and NO concentrations. A sensitivity analysis for the Mendiara-Glarborg mechanism indicated that remaining uncertainties of the rate constants implemented in the recent H/N/O models are difficult to scrutinize unambiguously due to experimental uncertainties.