The broadened concept of consumer behaviour is particularly important for residential electricity modelling. During the last years different time series approaches based on timing of electricity use and data on dwellers behaviour have been used in connection with residential electricity modelling for predicting the future demand. In this paper Markov chain modelling is used for simulation of activity patterns of households. Face-to-face interviews of 30 Latvian households were conducted in November, 2014 with the aim to retrieve more accurate data on user activities, in particularly, to find out whether and when the resident is active in the household each hour during the day. Markov chain transition probabilities were generated based on two time – varying Markov states: 1 – „active and using electricity” and 0 – „inactive and not using electricity”. The behaviour of the dwellers was differentiated among 4 typical daily modes: on weekdays and weekends in summer and winter. The modelling results showed that Markov chain simulation model creates a realistic coverage of activities over time and makes it possible to realistically reproduce consumption for the whole day. Statistical validation of the predicted electricity consumption (Markov chain model) confirms a close coincidence against extensive empirical time-use data (smart meter data) with high precision.