The aims of this study were to compare inter-observer variability within and between two specialized infant lung function testing centres and to develop a strategy for performing and analysing infant respiratory function tests to facilitate future collaborative trials. A protocol for data collection and analysis was developed using similar equipment and identical software. All raw data were exchanged on disk and analysed, blind to infant status. All data were cross-analysed by both centres to assess inter-observer variability. Outcome measures were functional residual capacity (FRCpleth), airway resistance (Raw) and maximal expiratory flow at FRC (V'max,FRC). Subjects were recruited from the multicentre UK extracorporeal membrane oxygenation (ECMO) Trial and measured at around 1 yr of age. Forty-two infants attended the Institute of Child Health, London and 36 attended the Leicester Royal Infirmary. The proportion of infants treated with ECMO or conventional management at each centre was similar. There were no significant differences between any of the outcome measures for infants tested at either centre. During a cross-analysis, the agreement between the two centres, within infant, was closer for V'max,FRC and FRCpleth (within 10%) than for the more variable measurements of Raw (within 20%). A collaborative approach to trials with infant respiratory function as an outcome measure appears feasible, providing that close attention is paid to study design, and participants in such trials maintain a standard approach to data collection and analysis.