For probabilistic interpretation and monitoring performance enhancement in noisy processes, the probabilistic principal component analysis (PPCA) method has recently been introduced into the monitoring area. However, PPCA is restricted in linear processes. This paper first gives a new interpretation of PPCA through the Gaussian process manner. Then a new nonlinear probabilistic monitoring method is proposed, which is developed upon the Gaussian process latent variable model. Different from the traditional PPCA method, the new approach can successfully extract the nonlinear relationship between process variables. Furthermore, it exhibits more detailed information of uncertainty for process data, through which the operation condition and the fault behavior can be interpreted more easily. Two case studies are provided to show the efficiency of the proposed method.