The purpose of this thesis work was to develop a headspace solid phase microextraction gas chromatography mass spectrometry (HS-SPME-GC-MS) method to detect volatile organic compounds (VOCs) in board samples and to statistically investigate potential correlation between chromatographic data and flavor data obtained from a trained panel. The developed method would hopefully serve as a complement to the already established routine analyses at Stora Enso and gain an increased understanding of which VOCs in the board influence its flavor properties. The impact of incubation time and adsorption time on the area under curve (AUC) was studied with a Design of Experiment screening using the software MODDE. The screening data showed a correlation between large AUC and low repeatability measured as relative standard deviation (RSD). The data was hard to fit to a model due to the large RSD values for the replicates, AUC for identified compounds as response gave an acceptable fit. The regression coefficients for the model showed that a longer adsorption time gave larger AUC, while incubation time had no significant impact on the response.  Instead of following up the screening with an optimization, the focus was shifted to improving the repeatability of the method, i.e. lowering the RSD. The high RSD was believed to mainly be the result of leakage of analytes and unstable temperature during adsorption, preventing the system from reaching equilibrium. Different heating options and capping options for the vial was tested. Septum in crimp cap ensured a gas tight seal for the vial, giving lower RSD values and larger AUC compared to the other alternatives, showing that there was indeed a leakage. Using oil bath ensured stable temperature during the adsorption and detection of a larger number of VOCs but created a temperature gradient in the vial due to it not being fully submerged in the oil. Oil bath gave larger AUC, but still high RSD due to the temperature gradient making the method sensitive to variance in fiber depth in the vial. The final method was performed with 2 g of board sample in a 20 ml headspace vial sealed with a crimp cap with septa. The incubation and adsorption were performed with the vial immersed in a 90-degree oil bath. 20 min incubation time was chosen based on the time it took to get a stable temperature gradient in the vial, and 20 minutes adsorption time was chosen as a good compromise between large AUC and low RSD. Compared to Stora Ensos routine analysis, the developed SPME method gave chromatograms with an improved signal-to-noise ratio for the base line and several more peaks with larger AUC. For the board sample used during method development, the SPME-method identified 34 VOCs, while the routine analysis only identified 12. The developed method was applied on 11 archived board samples of the same quality that were selected based on their original flavor properties, to get a large diversity of samples. Flavor analysis was performed by letting a trained flavor panel describe the flavor based on intensity and character of the water that had individually been in indirect contact with one of the 11 board sample for 24 h. Potential correlation between chromatographic data obtained with the developed method and the flavor experience described by the flavor panelists was statistically investigated with the multivariate analysis software SIMCA. The correlation study showed that a combination of 12 VOCs with short retention time are most likely the main source of off-flavor which of 5 could only be identified with the developed SPME method. VOCs with long retention time did not contribute to an off-flavor and might have a masking effect on flavor given by other VOCS, however not confirmed in this study. Furthermore, the age of the board samples proved to be a good indicator for prediction of the flavor intensity, whereas the total AUC of the samples was not. Possible correlation between detected VOCs in the samples and flavor character given by the flavor panel were seen, however the variation in the data and the sample set were too small, preventing from making conclusions on individual VOCs impact on the flavor experience. The developed HS-SPME-GC-MS method would serve as a complement to the already established routine analyses at Stora Enso and has slightly increased the understanding of which VOCs in the board influence the flavor properties