Cryptococcus neoformans is an encapsulated fungus that widely exists in the environment through the world. It can enter the human body through the respiratory tract, causing inflammation of the lungs and even causing serious infectious diseases in patients with impaired immune function. Those serious infectious diseases can be avoided if the cryptococcal pneumonia can be early detected and accurately assessed. However, due to less clinical symptoms, cryptococcal pneumonia is easily misdiagnosed as lung cancer, tuberculosis, etc., and the rate of misdiagnosis is high. Therefore, a rapid, accurate and non-invasive method is urgently needed to make early diagnosis and precise assessment of cryptococcal pneumonia, in which timely and appropriate treatment can be implemented and the cure rate is expected to be significantly improved. In this study, surface enhanced Raman spectroscopy (SERS) was served as the diagnostic tool to identify cryptococcal pneumonia, in which the blood serum SERS spectra are collected from cryptococcal pneumonia mice and uninfected mice. The multivariate curve resolution alternating least squares (MCR-ALS) and support vector machine (SVM) are employed to quantify the biochemical composition changes in serum microenvironments. The excellent results have demonstrated that the serum SERS has significant potential in early diagnosis and accurate assessment of cryptococcal pneumonia.