Abstract: A medical claims analysis system and method categorizes medical claims into episodes of care having predetermined diagnostic cluster types. The system analyzes medical claim items, some of which have principal diagnosis codes, and some of which have non-principal, missing, or incorrect diagnosis codes. Patient treatment episodes (PTEs) are formed from the principal diagnosis codes, each PTE being of a particular diagnostic cluster type. The system categorizes non-principal-diagnosis claim items into the PTEs on the basis of temporal, physiological or clinical relationships between the claim items and the PTEs. A drug lookup table enables drug claims to be properly categorized in the PTEs. A diagnostic cluster lookup table enables claim items to be categorized into PTEs with ongoing treatment windows for which the diagnosis code of the claim item is in the diagnostic cluster lookup table. The system merges PTEs of the same diagnostic cluster type when the treatment windows of the PTEs overlap.