Behavioural complexity is likely to reflect how animals cope with their environment. A large behavioural repertoire and the ability to flexibly apply these behaviours provide an individual with a greater likelihood of adapting to its (captive) environment. Here, we developed a procedure to aggregate different behavioural measures into a condensed measure of behavioural complexity based on 14 features, which were previously proposed (e.g. number of behaviours, Shannon diversity index) as well as some new features of behavioural complexity (e.g. variances of within and between transition durations). To test the measure, artificial behavioural sequences with potentially varying complexity were created using an individual-based modelling approach. With a Principal Component Analysis, the features extracted from these sequences could be reduced to two components (‘general complexity’ and ‘state variability’) explaining 59.64 and 27.68% of the total variance, respectively. The effect of the aspects of the artificial behavioural sequences on ‘general complexity’ and ‘transitions variability’ were analysed using linear mixed-effects models. The number of behavioural categories and the proportion of short behavioural states had the largest effect on the components. Both components were increasing with a greater number of behavioural categories, whereas the proportion of short behavioural states the opposite effect on the components. We also tested the approach on real data-sets. The principle components were not identical to the ones from the simulation. Yet, we found consistencies and similarities in the loadings. The approach for an aggregated measure of behavioural complexity developed here could form the basis of an individual-based animal welfare indicator for intensively kept animals.