This study estimates a disaggregate freight generation model with the French shipper survey ECHO. This survey contains accurate information about French shippers, with variables describing their economic activity, the diversity of their production, their relationships with their clients and carriers, etc. These variables and their influence on production and attraction are first investigated sequentially. Then generation models are estimated using ordinary least squares, through various approaches: continuous explanatory variables only, continuous and qualitative variables and their interactions, and finally simple models for applications with limited data availability.