The objective of this study was to explore methods examining patient complexity and therapy interventions in relation to functional outcomes from an inpatient multiple sclerosis (MS) rehabilitation program. Retrospective and prospective data for 24 consecutive inpatients at a tertiary rehabilitation facility assessed (i) impairment/disability--Expanded Disability Status Scale, Neurological Impairment Set, (ii) complexity of intervention--Rehabilitation Complexity Scale, Northwick Park Therapy Dependency Assessment (NPTDA), the type and duration of therapy, (iii) Function-Barthel Index (BI), functional independence measure (FIM), (iv) hospital length of stay, discharge destination. The mean length of stay was 20 days (SD=15.7); male:female ratio was 10:14; 11 (46%) had Expanded Disability Status Scale scores more than 6.5, and all discharged to community. Over one-half had "moderate-severe" scores for fatigue, and deficits in motor function and mood that resulted in significant functional limitation. Rehabilitation Complexity Scale scores showed substantial complexity with two-thirds requiring specialized nursing. The NPTDA score (median 22 interquartile range 20-23) showed moderate dependency on admission in the following domains: physical, cognitive and psychosocial programs and preparing for discharge. The mean physiotherapy and occupational therapy received was 45.76 and 24.04 min/day, respectively. Functional gains from admission to discharge for FIM and BI were significant (P less than 0.001). There were strong correlations between the total NPTDA and FIM motor scores (Spearman's [rho] - 0.80) and BI ([rho] - 0.83), but only weak correlations with the FIM cognitive score ([rho] - 0.33). In conclusion, this pilot study provided information about the complex interplay of patient and process factors, and their interrelationships that impact functional outcomes in MS rehabilitation. A prospective study is now planned using appropriate tools to understand the "black box" of rehabilitation.