The lithium in serum of the patients should be monitored to achieve the regular medical effect and prevent from lithium poisoning when the medicine containing lithium is adopted to treat the manic depression. At present, flame atomic absorption spectrometry and electrochemical methods are commonly used in lithium determination in clinic area. However, for the flame atomic absorption spectrometry method, the instrument can’t be portable as the C2H2 cylinder is necessary, for which the demands for the laboratory are critical. As for the electrochemical technique, the treatment of the electrodes is tedious and the analytical efficiency is somewhat low. The solution cathode glow discharge-atomic emission spectrometry, which has proved to be sensitive for Li determination with simple operation and portable instrument, has never been used in clinic Li test in serum. Therefore, a novel method was established for sensitive determination of lithium in serum based on a home-made portable solution cathode glow discharge- atomic emission spectrometer. The serum samples were pumped into the discharging chamber after dilution and acidification. Lithium in serum was excited by the plasma generated between the solution cathode and the solid metal anode, and the characteristic spectrum came into being in the common atmosphere and the signal collection and analysis were carried out by using a fiber optical spectrometer. The influence of dilution times within 10～100 on matrix interferences was comprehensively investigated and the results showed that no significant matrix interferences were observed when the dilution times were above 20 and the reliability could be basically ensured. The parameters influencing the analytical performances, including the analytical emission line, the acidity of the testing solution, etc, were optimized. The results showed that 671 nm was proved to be the ideal analytical emission line due to the strongest emission intensity and the non-significant spectral interferences while the best analytical performance could be achieved when 1% HNO3（V/V）was adopted to keep the acidity of the solution on the premise of the suitable stability. Under the optimized conditions, the equation of the linear calibration curve was as follows: Ie=7 299ρLi+400; R2=0.998 3. The limit of determination for lithium in the serum was 0.2 mg·L-1 when the serum sample was diluted by 20 times. The relative standard deviation for the repeatability test based on 6 times tests of the serum was below 5%. This method was applied to the analysis of serum standard material. The detection result was in favorable agreement with the certified value, which showed that this method can be used in the determination of lithium in real serum samples of clinical tests.