In this work, multiple extraction thermal desorption (METD), as a sample introduction method for GC, was developed. This technique was used for the determination of residual solvents (RS) in modified cellulose, because it is practically impossible to dissolve or distribute it uniformly in water and common organic solvents. Moreover, METD facilitates the optimization of the desorption time and it is more sensitive to quantify trace level volatiles in insoluble material, compared to direct dynamic desorption (DDD). In addition, METD provides diagnostic information about the sample-sorbent interaction. Three solvents (methanol, ethanol and tert-butanol) were determined in two types of modified cellulose (dialdehyde cellulose (DAC) and DAC-ethylenediamine (DAC-EDA)). It was shown that good linearity over a wide concentration range was achieved. The limits of detection (LOD) and limits of quantification (LOQ) for the different solvents ranged from 0.1 to 0.3 μg and from 0.3 to 0.9 μg per tube, respectively. Accuracy of the METD method was verified by using an alternative method based on the decomposition of the modified celluloses by Trichoderma reesei cellulase, followed by headspace-trap-GC (HS-trap-GC). The results obtained from the two validated methods were found to be similar (relative deviation < 17.0 %). However, the developed METD-GC method is preferable for the analysis of RS in modified cellulose since it does not require sample pretreatment and possesses higher sensitivity.