Three inter-related studies examine the construct of problem solving as it relates to the assessment of deficits in higher level outpatients with traumatic brain injury (TBI). Sixty-one persons with TBI and 58 uninjured participants completed measures of problem solving and conceptually related constructs, which included neuropsychological tests, self-report inventories, and roleplayed scenarios. In Study I, TBI and control groups performed with no significant differences on measures of memory, reasoning, and executive function, but medium to large between-group differences were found on timed attention tasks. The largest between-group differences were found on psychosocial and problem-solving self-report inventories. In Study II, significant-other (SO) ratings of patient functioning were consistent with patient self-report, and for both self-report and SO ratings of patient problem solving, there was a theoretically meaningful pattern of correlations with timed attention tasks. In Study III, a combination of self-report inventories that accurately distinguished between participants with and without TBI, even when cognitive tests scores were in the normal range, was determined. The findings reflect intrinsic differences in measurement approaches to the construct of problem solving and suggest the importance of using a multidimensional approach to assessment.