Acute respiratory distress syndrome (ARDS) is a highly heterogeneous syndrome with a continuing high mortality rate. Despite intensive research, established therapies consist mainly of supportive measures, while pharmacological approaches have not yet shown any consistent survival benefits. In recent years, it has become clear that the great clinical and biological diversity of ARDS contributes significantly to the difficulty of demonstrating therapeutic effects. The phenotyping of ARDS has therefore become a central field of research. Different approaches—from clinical parameters and imaging to inflammatory and cardiovascular profiles and multi-omics analyses—have repeatedly identified reproducible subphenotypes that differ in prognosis and, in some cases, in response to therapies. Hypo- and hyperinflammatory subphenotypes have been described as particularly consistent. These are prognostically relevant and, in retrospective analyses, have also shown a differentiated response to glucocorticoids, statins, or fluid strategies. However, endotypes based on causal pathophysiological mechanisms are still largely theoretical. The concept of treatable traits illustrates the potential of personalized therapy but is currently based predominantly on retrospective findings. Future studies should use standardized terminology and multimodal approaches, take longitudinal data into account, and aim for prospective validation to define robust subphenotypes and causal endotypes. This could lay the foundation for true precision medicine in ARDS.