Digitalization in the oil and gas industry has led to the formation of digital twins. Digital twins bring closer the physical and virtual world as data is transmitted seamlessly between real time sensors, databases and models. The strength of the digital twin concept is the interconnectivity of data and models. Any model can use any combination of inputs (e.g. operator owned data sets and sensors, third-party databases such as soil composition or weather data, results from other models such as flow assurance, threat modelling or risk modelling). Consequently, the result of one model may become the input of another. This strength is also a weakness, as uncertain (or missing data) will lead to a great source of uncertainty and may lead to wrong results. Worst case scenarios have been used to solve this issue without success. This paper presents a new concept: probabilistic digital twins for pipelines. Probabilistic digital twins do not lose uncertainty as results pass from one model to another, thus providing greater confidence in the final results. This publication reviews the probabilistic digital twin concept and demonstrates how it can be implemented using gas pipeline data from West Pipeline Company, CNPC.