Purpose In search of Kipling’s six honest serving men in upper limb rehabilitation after stroke, we sought to investigate clinicians’ perspective of when and where to begin therapy, how much and what therapy to provide, and who and why (or not) to provide therapy. Materials & methods: Within-participant case cross-over experiments were nested within an anonymous web-based questionnaire (21 questions, three cases). Graph theory-based voting to produce ranked ordered lists and mixed-effect logistic regression were performed. Results In total, 225 Australian stroke clinicians responded: 53% occupational therapists, 61% working in acute/inpatient stroke setting. Most respondents indicated they did not have a protocol/expectation regarding when (62%), how much (84%) or what (60%) therapy to provide in their setting. Respondents ranked 24-h to 7-days post-stroke as the optimal time to commence therapy, and 30- to 60-min per day as the optimal dose to provide. Within-participant experiments demonstrated that greater motor recovery as time progressed increased the odds of offering therapy, while lack of motor recovery, shoulder pain, neurological decline and sole therapist reduced the odds. Conclusion We need to develop an evidence base concerning Kipling’s six honest serving men and equip clinicians with clinical decision-making skills aligned with this focus. IMPLICATIONS FOR REHABILITATION Most clinicians did not have access to a protocol / clinical pathway which defines when, how much and what upper limb therapy to provide after stroke, which may be improved by providing individual clinicians with organisational support to make therapy decisions. To improve the personalisation of upper limb rehabilitation in clinical practice, we need to understand when and where after stroke to begin therapy, how much and what therapy to provide, as well as who and why (clinical decision-making) to provide therapy. Clinicians perceive clinical trials as successful if the therapy can demonstrate recovery that is greater than a minimal clinical important difference (MCID).