OBJECTIVES:Previous studies found that hospital and specialty have limited influence on patient experience scores, and patient level factors are more important. This could be due to heterogeneity of experience delivery across subunits within organisations. We aimed to determine whether organisation level factors have greater impact if scores for the same subspecialty microsystem are analysed in each hospital. SETTING:Acute medical admission units in all NHS Acute Trusts in England. PARTICIPANTS:We analysed patient experience data from the English Adult Inpatient Survey which is administered to 850 patients annually in each acute NHS Trusts in England. We selected all 8753 patients who returned the survey and who were emergency medical admissions and stayed in their admission unit for 1-2 nights, so as to isolate the experience delivered during the acute admission process. PRIMARY AND SECONDARY OUTCOME MEASURES:We used multilevel logistic regression to determine the apportioned influence of host organisation and of organisation level factors (size and teaching status), and patient level factors (demographics, presence of long-term conditions and disabilities). We selected 'being treated with respect and dignity' and 'pain control' as primary outcome parameters. Other Picker Domain question scores were analysed as secondary parameters. RESULTS:The proportion of overall variance attributable at organisational level was small; 0.5% (NS) for respect and dignity, 0.4% (NS) for pain control. Long-standing conditions and consequent disabilities were associated with low scores. Other item scores also showed that most influence was from patient level factors. CONCLUSIONS:When a single microsystem, the acute medical admission process, is isolated, variance in experience scores is mainly explainable by patient level factors with limited organisational level influence. This has implications for the use of generic patient experience surveys for comparison between Trusts and should prompt further research to explore if more discriminant surveys can be developed.