In this paper, we address the important issue of pathways of protein folding. With the statistical energy landscape theory, the folding can be seen as a chemical reaction process from an ensemble of unfolded states to a unique folded state. The Poissoness (non-Poissoness) of statistics of kinetic populations of unfolded states and folded state can be related to the multiple (discrete) pathways of protein folding. In large populations, the statistical fluctuations are suppressed by the number of molecules in the events. In single molecule experiments, fluctuations in general cannot be ignored and different statistics becomes prominent and detectable. From the analytical study, we find that multiple pathways emerge when the bias towards the folded state is large compared to the fluctuations of the energy landscape funnel and the kinetic population is single exponential in time indicating a Poisson process. On the other hand, when fluctuations or roughness of the landscape dominates, discrete pathways of folding emerge and the kinetic population exhibits nonexponential behavior in time indicating a non-Poisson process. The advances in single molecule detection may provide a good opportunity and a new way of statistical analysis of the kinetic data to resolve the multiple pathway versus discrete pathway issue of protein folding. (C) 2003 American Institute of Physics.