During the process of biomass gasification tars are formed which exit the gasifier in vapor phase. Tar condensation creates problems like fouling and plugging of after-treatment, conversion and end-use equipment. Gasification tars consist mainly of Polycyclic Aromatic Hydrocarbons (PAHs). Former research has shown the possibilities and difficulties of tar conversion by partial combustion. Basic studies to investigate the oxidation of tars in non-premixed combustion processes are expected to give more insight in this problem. In this paper the ability of the flamelet-generated manifold (FGM) approach to numerically model multi-dimensional, laminar, non-premixed flames with the inclusion of PAH chemistry is investigated. Modeling detailed PAH chemistry requires the employment of large reaction mechanisms which lead to expensive numerical calculations. The application of a reduction technique like FGM leads to a considerable decrease (up to two orders) in the required computation time. A 1D numerical validation shows that the improvements achieved by implementing a varying Lewis number for the progress variable Y are significant for PAH species with a large Lewis number, such as C10H8. Considerable improvements are found near the flame front and on the fuel side of the flame. A comparison has been made of FGM results with qualitative Planar Laser Induced Fluorescence (PLIF) measurements. A laminar CH4/N2–air co-flow flame has been doped with two dopants, benzene and toluene, at three different concentrations. A set of filters was used in order to qualitatively distinguish the small (1–2 rings) and large (3 or more rings) aromatic species. The results show that the model is able to capture the major flame characteristics typical for PAH formation in multi-dimensional laminar non-premixed flames.