In the context of growing concerns over energy consumption and sustainability, accurate modelling of occupancy patterns within residential buildings is critical. In this study, a novel stochastic occupancy model is introduced for simulating human behaviour within residential buildings by employing Time Use Survey (TUS) data and utilising Markov chains and probabilistic sampling algorithms. The novelty of this research lies in its approach to represent the dynamic nature of occupancy across different functional spaces and age groups, a gap not yet adequately addressed in existing studies. The model's accuracy is ascertained through ten-fold cross-validation, achieving an average R2 value of 0.91 across key functional rooms (bedroom, bathroom, kitchen, living room), indicating a high degree of precision. Applied to a case study of a two-story detached house in the UK, the model effectively reflects varied behaviour patterns and room occupancy among different age groups. For instance, the average daily appliance energy consumption for occupants aged 8–14 ranged from 0 to 3.77 kWh (median 1.71 kWh), for ages 15–64 from 0 to 4.93 kWh (median 2.61 kWh), and for over 65 from 0.87 to 5.65 kWh (median 3.60 kWh). This model, with its scalability and accuracy in capturing the inherent randomness of human behaviour, is a valuable tool for improving energy consumption simulations and contributing to sustainable residential building design and management.