Drug levels in patients' bloodstreams vary among individuals and consequently therapeutic drug monitoring (TDM) is fundamental to controlling the effective therapeutic range. For TDM purposes, different analytical approaches have been used, mainly based on immunoassay, liquid chromatography- ultraviolet, liquid chromatography-mass spectrometry and liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods. More recently a matrix-assisted laser desorption/ionisation method has been proposed for the determination of irinotecan levels in the plasma of subjects under therapy and this method has been cross- validated by comparison with data achieved by LC-MS/MS. However, to reach an effective point-of-care monitoring of plasma drug concentrations, a TDM platform technology for fast, accurate, low-cost assays is required. In this frame, recently the use of paper spray mass spectrometry, which is becoming a popular and widely employed MS method, has been proposed. In this paper we report the results obtained by the development of a paper spray-based method for quantitative analysis in plasma samples of imatinib, a new generation of anticancer drug. Preliminary experiments showed that poor sensitivity, reproducibility and linear response were obtained by the "classical" paper spray set-up. In order to achieve better results, it was thought of interest to operate in presence of a higher and more homogeneous electrical field. For this aim, a stainless steel needle connected with the high voltage power supply was mounted below the paper triangle. Furthermore, in order to obtain valid quantitative data, we analysed the role of the different equilibria participating to the phenomena occurring in paper spray experiments, depending either on instrumental parameters or on the chemical nature of analyte and solvents. A calibration curve was obtained by spiking plasma samples containing different amounts of imatinib (1) with known amounts of deuterated imatinib (1d3) as internal standard, with molar ratios [1]/[1d3] in the range 0.00-2.00. A quite good linearity was obtained (R2 = 0.975) and some experiments performed on spiked plasma samples with known amounts of 1 confirmed the validity of this method.