Complexity has become an increasingly prominent concept in research on mental disorders and pathological behaviors. This is more than due, as these conditions are not only complicated but genuinely complex. Complexity serves both as an analytic lens and as a force that pushes scientific progress forward. However, complexity science, as it currently stands, seems poised to deliver waves of new findings but no transformative breakthroughs. We propose rethinking complexity from a multisystem perspective which has been common in social sciences and naturally fits with the setting in mental and behavioral disorders, respectively. The key feature of multisystem complexity is heterogeneity of involved systems, heterogeneity of involved processes, their interactions and outcomes. Implementing a multisystem and heterogeneity perspective requires a thorough rethinking of methodology, while assessing failures, misunderstandings, requirements and prospects. The same is true for research directions such as risk factor research. Examples of implementations are drawn from research on suicide and neurodevelopmental disorders.