Definitive analysis of solute concentrations in lung lavage fluid involves the use of a marker of dilution to correct for variable recovery of epithelial lining fluid (ELF), but the question of the most appropriate dilutional marker remains unresolved. In lavage fluid collected from infants with lung disease and healthy control subjects, we examined ELF concentration of protein, albumin, sphingomyelin (SM), and IgA secretory component (SC), and critically appraised the relative validity of SC and urea as dilutional markers in the context of lung infection and lung injury. Protein, albumin, and SM were found not to be valid dilutional markers, as their ELF concentration varied significantly between the diseased, recovering, and normal lung. Differences in concentration were noted in both tracheal aspirate samples (TA, 4 x 0.5 ml) and nonbronchoscopic bronchoalveolar lavage fluid (NB-BAL, 3 x 1 ml/kg), but were not uniform (e.g., TA-disease versus control: albumin 2.8 versus 0.68 mg/ml, SM 45 versus 16 microgram/ml, both p < 0.05; NB-BAL-disease versus recovery: protein 8.1 versus 4.8 mg/ml, albumin 2.9 versus 1. 4 mg/ml, both p < 0.05). Overall, SC concentrations in ELF were not different between the diseased and normal lung, but in the NB-BAL samples, significantly higher SC concentration was noted in viral bronchiolitis and pneumonia than in noninfective lung diseases. No clear evidence of additional influx of urea into lavage fluid in association with epithelial disruption was found in the diseased lung. Comparative analysis of SC and urea revealed no difference in TA samples, but in NB-BAL specimens, urea best standardized the lavage concentration of surfactant indices to correspond to the degree of lung dysfunction as indicated by oxygenation index. We conclude that SC and urea, but not protein, albumin, or SM, are valid dilutional markers with which to estimate ELF recovery during small volume lung lavage. Urea appears a more appropriate choice in return fluid derived from the distal tracheobronchial tree, and SC should not be used in the context of lung infection.