Load flow studies are an essential task carried out in power system planning and operations. However, the widely used deterministic load flow analysis is limited in their handling of network uncertainties or inaccuracies in input data. Therefore, probabilistic load flow (PLF) based on either numerical methods (such as Monte Carlo simulation) or an analytical method (such as convolution techniques) was developed in the 1970s to handle power system uncertainties due to variations in electrical network variables. This paper presents a probabilistic load flow analysis method based on convolution techniques. The method is suitable for distribution systems and examines the effect of load, generation and network uncertainties either separately or in combinations thereof. The main features on the paper include: a critical appraisal of existing PLF techniques, as published in the literature, is carried out to derive the optimised technique and methodology. The proposed method is applied to practical 47-bus radial distribution network modelled in PowerFactory DIgSILENT software package. The results obtained are then exported into MATLAB for detailed statistical analysis in terms of various probability distribution function (PDF) and cumulative density functions (CDF).