Purpose– User-oriented care, defined as individualized assisting behaviors, is the dominant approach within elderly care today. Yet, there is little known about its conceptual structure. This paper proposes that user-oriented care has a bi-partite structure which may be decomposed into the two dimensions of task and relation.Design/methodology/approach– Care workers were “shadowed” (i.e. observed) at their work (n=391 rated interactions). User-oriented care was assessed along ten process quality indicators targeting the acts of caregiving (i.e. task focus, relation focus, involvement, time-use, body language, autonomy, respect, warmth, encouragement, and information) in two elderly care settings, i.e. home care and nursing home. Observations added up to 45 hours.Findings– Principal component analyses confirmed the proposed two-factor structure of user-oriented care. Specifically, the user-oriented care indicators loaded on two distinct factors, i.e. task and relation. The underlying structure of user-oriented care revealed to be invariant across the two settings. However, the results revealed interesting structural differences in terms of explained variance and the magnitude of factor loadings in the home care and nursing home settings. Differences also emerged specifically pertaining to the indicators of autonomy and time-use. These findings suggest that user-oriented behavior may to some extent denote different acts of caregiving and what may be called task- and relation-orientation may be loaded with different meanings in these two care settings.Originality/value– This is the first study investigating user-oriented behavior in the context of elderly care using a quantitative observational approach. The authors propose that the observed differences between the two care settings are primarily not due to better elderly care work in home care, but due to some inherent differences between these two contexts of care (e.g. better health and living at home).