Birthweight is a key predictor of future life outcomes. When this information is not available we must rely on maternal recall of weight or size. We evaluated the use of maternal recall of size as an indicator for birthweight, using data from the 2004 Demographic and Health Survey in Cameroon. The overall agreement between recorded and recalled weight and subjective assessment of birth size was 90% (Kappa = 0.46). Concordant descriptions were associated with higher birthweights (P < 0.001), higher education (P = 0.008) and delivery in a health unit (P = 0.025). Specificity for low birthweight (92.9%) was much higher than sensitivity (59.9%) and the negative predictive value (96.1%) was much higher than the positive predictive value (44.4%). These results indicate that recall of size, in Cameroonian women and in other low resource settings, should be used only in the absence of other sources of data.