SummaryA series of attempts to develop a behavioural classification system for use in a first stage analysis of infant vocalization data are reported. The data were cross‐sectional and obtained from 11 different infants between the ages of 1 and 40 weeks. A set of terms and definitions were developed by two judges and their reliability assessed by having a third judge classify the data, using the terms and definitions in written form only. Implifications of the findings with respect to inter‐judge agreement and disagreement and usefulness of system for further studies of infant vocalization are discus