This work presents a mid-IR direct tunable diode laser absorption spectroscopy (dTDLAS)-based HCl spectrometer, which is specially designed and optimized to measure HCl concentration in combustion exhaust gas matrices (i.e. elevated gas temperatures, high water vapour and CO2 contents). The work is motivated by (legal) requirements for monitoring combustion emissions from large-scale power stations or biomass burning domestic boilers. In our novel dTDLAS HCl spectrometer we use a low power room temperature mid-IR ICL diode laser to access the HCl P5 line at 2775.76 cm−1 in the 1-0 vibrational band which was especially selected for gas matrixes with high CO2 and H2O admixtures. With this set-up we demonstrate at 77 cm path length, total pressure from 320 hPa to 954 hPa, room temperature and at 110 s of signal averaging an optimal precision of 0.17 µmol/mol (ppm). Gas monitoring in combustion applications and elevated gas temperatures are prone to systematic errors caused by spectroscopic falsifications in colder gas boundary layers (BL) unavoidable in the high temperature gas ducts. These BL lead, e.g. to temperature, matrix composition or target gas concentration gradients near walls, which also influence the spectroscopic raw signal via their temperature and collision partner dependence. Depending on the chosen spectral line these can cause significant systematic deviations in in-situ, line-of-sight (LOS) laser spectrometers. For an improved understanding of the quantitative effects of thermal BL on our LOS HCl dTDLAS spectrometer and better representing real BL, we expanded our spectroscopic BL simulation model to allow for continuous (linear) instead of the previous stepwise changes. From the new simulation results, we deduce systematic relative deviations in the extracted HCl-concentration to be up to 10% depending on the magnitude of the BL changes and the choice of the “representative” single temperature measurement. With this simulation model, a user can now derive the system’s systematic deviation based on assumptions on the present temperature gradients. The model also helps the user to choose the gas temperature measurement location in his process, in order to minimize the spatial heterogeneity effects.