Aim We propose a higher order construct termed Employee Social Liability (ESL), incorporating four clusters of employee “bad” behaviour and integrating them under a common latent variable. The key contribution of this paper is to create, describe and validate a measure of ESL to aid in the study of negative relationships at work. Design To develop our measure of social liability, we conducted a series of linked, qualitative and quantitative studies. The overall research design being mixed-methods. Method Study 1 focused on generating a pool of potential items to measure the theorised four dimensions of the social liability construct. Study 2 utilised a Q-sort process to systematically reduce this item pool. Study 3 refined this reduced pool of items by analysing their factor structure and other psychometric properties in order to examine their measurement of employee behaviour as it contributes to employee social liabilities. 1000 participants completed an online questionnaire containing the 42 items retained from the previous two studies along with measures of other aspects of work and also individual differences including deviant work behaviours, personality and EQ. Validation of the measure and the dimensionality of the social liability items was undertaken by estimating a series confirmatory factor analysis models using AMOS (V22). This final study also examined the nomological network of ESL with these other variables. Results We provide empirical evidence that ESL represents a higher-order construct incorporating four categories of employee bad behaviour in the domains of 1) experiencing distrust, 2) lack of cooperation, 3) social demand, and 4) negative relationships. Finally we provide a nomological network for the ESL construct. Conclusion We provide initial evidence that the negative relational constructs distrust, lack of reciprocity, negative interactions and distraction have a common core labelled Employee Social Liabilities. Further we provide evidence that ESL can be measured and that it is related to one’s own bad workplace behaviours, low emotional intelligence and inversely to having supportive colleagues. We anticipate that this measure will prove useful to the study of behaviour at work, and will facilitate further empirical research into this important area of enquiry.