The aim of this two-wave study is to investigate whether burnout, work engagement and workaholism can be empirically distinguished in one model and whether this model shows structural stability over a period of 2 years (i.e. whether the distinguishability between the constructs holds across time). The study was conducted among 118 judges in the Netherlands who completed questionnaires measuring burnout, work engagement and workaholism. The results showed that these are relatively distinguishable constructs, despite a considerable overlap of professional efficacy loading on work engagement (instead of burnout; as hypothesized), absorption loading on workaholism (in addition to work engagement; as hypothesized) and exhaustion loading on workaholism (in addition to burnout), which represents a new finding. These extra loadings led to model modifications, which were found at both time points. As hypothesized, this model appeared to be stable over time. Nevertheless, further clarification and conceptualization of these constructs are undoubtedly needed.