Picture naming has become an important experimental paradigm in cognitive psychology. Young children are more variable than adults in their naming responses and less likely to know the object or its name. A consequence is that the interpretation of the two classical measures used by Snodgrass and Vanderwart (1980) for scoring name agreement in adults (the percentage of agreement, based on modal name, and the H statistic, based on alternative names) will differ because of the high rate of "don't know object" responses, common in young children, relative to the low rate of "don't know object" responses more characteristic of adults. The present study focused on this methodological issue in young French children (3-8 years old), using a set of 145 Snodgrass-Vanderwart pictures. Our results indicate that the percentage of agreement based on the expected name is a better measure of picture-naming performance than are the commonly used measures. The norms may be downloaded from www.psychonomic.org/archive.