In a retrospective study, urinary malondialdehyde concentration in 45 preterm neonates (25-35 weeks' gestation) during their first month of life was measured by HPLC. Urine was collected on different days of life as a 3-h sample. The frequency of urine collection and measurement varied between one (n = 22) and seven times (n = 8) per child. The study group was divided into three categories according to birth weight: low-birth-weight infants (LBW) (n = 16), very low-birth-weight infants (VLBW) (n = 17) and extremely low-birth-weight infants (ELBW) (n = 12). Urinary malondialdehyde concentration was highest in the ELBW group: 1. 15 (0.66, 2.12) mumol/l (median and quartiles) versus 0. 58 (0.34, 1.18) mumol/l in the VLBW and 0.60 (0.40, 1.06) mumol/l in the LBW groups (ELBW versus VLBW, p < 0.005; ELBW versus LBW, p < 0.02). In oxygen-treated neonates, significantly higher malondialdehyde values were found compared to those without supplementary oxygen (0.89 (0.48, 1.74) versus 0.58 (0.32, 0.89), mumol/l; p < 0.005). Likewise, a higher malondialdehyde concentration was found in infants requiring mechanical ventilation (intermittent mandatory IMV or high frequency ventilation) compared to those breathing spontaneously (intermittent mandatory ventilation: 0.80 (0.42, 1.66); p > 0.05 and high frequency ventilation: 1.20 (0.83, 2.13); p < 0.001 versus 0.57 (0.33, 0.88) mumol/l). Malondialdehyde concentrations correlated significantly with FiO2 values of the individual patients (r = 0.22; p<0.02). Comparing urinary malondialdehyde concentrations in infants with and without bronchopulmonary dysplasia, a significantly higher malondialdehyde concentration was found in the former group (0.96 (0.51, 2.07) versus 0.60 (0.32, 0.98) mumol/l; p < 0.005)). Infants with patent ductus arteriosus had a higher urinary malondialdehyde concentration than those without patent ductus arteriosus (1.04 (0.58, 3.78) versus 0.64 (0.36, 1.20), mumol/l; p > 0.05)). Malondialdehyde concentrations were also higher in infants with intracranial bleeding compared to those without (0.83 (0.46, 1.42) versus 0.56 (0.33, 1.10) mumol/l; p < 0.02)). No significant differences in urinary malondialdehyde concentration were seen, either in relation to iv feeding with or without lipid emulsion or to medication administered. Native malondialdehyde concentration in seven commercial preparations of lipid emulsion after various periods of storage was fairly constant (12.3 +/- 0.4 mumol/l) (mean +/- SD).