The objective specification of the second variable (body weight) measured in an epidemiological field study of weanling diarrhoeal disease of infants has been approached by computer-based pattern analysis, complementing previous studies of the diarrhoeal signal of daily defaecation count (DS). The weight data are available for 64 weeks in a subset of 50 out of 139 infants in the study. A weekly incremental weight signal (IWE) was estimated by successively differencing each new observed weight with a smoothed version of the previous value. Variable latency of the different features of this variable precludes the formation of an average-profile by coherent averaging the ensemble of records; a feature-analysis method based on statistical trend analysis was therefore developed. Features of the resulting IWE average profile can be linked to weanling events, transient diarrhoeal episodes, and typical post-neonatal shifts in level of the diarrhoeal signal. The results are consistent with the view that both nutritional and infectious factors operate in most deviations from typical behaviour seen in individual IWE records. In all the infants having an abnormally low (64 week) weight gain, a sustained abnormal DS elevation can be recognized in the 7-15 week period, with recurrent diarrhoeal episodes (type P2, as previously classified) during the 45-60 week period; both abnormalities were associated with substantial effects on weight gain. But individual P2-diarrhoeal episodes occurring before about 40 weeks of age could be linked only to small, quickly-compensated effects on incremental weight.