This study assesses psychometric and empirical characteristics of the Multi-Attribute Health Status Classification (MAHSC) in a heterogeneous population of children admitted to intensive care. The MAHSC encompasses six health domains (sensation, mobility, emotion, cognition, self-care, and pain), each with four or five hierarchic levels of dysfunction. The health status of 254 consecutively admitted children was determined independently by 3 different observers (parents, attending clinicians, and investigators). The proportion of children with health impairment varied between 13% (in sensation) and 58.7% (in mobility). Interrater reliability of domain scores, as analyzed by Spearman's rank, Pearson product moment, and intraclass correlations, was high, generally exceeding 0.80 for all pairs of observers. Intraclass correlation appeared to be highest in nonsurgical patients (range, 0.89-0.98) and lowest in cardiovascular surgery patients (range, 0.50-0.84). The validity of the classification was supported by (1) the presence of expected specific health status impairment in patient suffering from diseases with an acknowledged impact on specific domains and (2) the presence of a relation between health impairment and medical consumption. The absence of a relation between the average (chronic) health status preceding admission and the acute risk of mortality precipitating the admission confirmed their independent prominence in outcome assessment. We conclude that the MAHSC is a feasible, reliable, and valid measure for outcome assessment in a heterogeneous population of children within a demanding clinical situation. Excellent interrater reliability allows the use of various raters, adjusted to practical requirements.