In order to be more competitive, aerospace industries have to continually improve product's cost, quality and cycle time. One way to perform all three at once is to manage product's variations, as variations are responsible for waste, rework and delay. There are only three ways to reduce product's variation: improve manufacturing capability, improve assembly capabilities and manage assembly sequence. Generally, variation reduction programs tackle only manufacturing capabilities. Unfortunately this solution is by far the most expensive one. We propose in this article to reduce variation first by selecting best assembly sequences or eliminating assembly tools and only after if there is no other way by reducing manufacturing capabilities. With this aim, we have defined an assembly analysis method. This method studies for each assembly sequence how variations are flowed from part to part through the assembly joints. Two types of assembly joint have been defined: mates that pass positional variation from part to part and contact that merely provide support and block variations. To represent variation flow down, we propose to use an “Assembly-Oriented Graph” and a “Datum Flow Chain”. First one captures the assembly sequence and allows us to identify which joints are mates and which ones are contacts. The second represents where variations are driven in the product. Datum Flow Chains assess assembly families and best ones are selected. This method can be used from the preliminary design allowing validating assembly sequence and design principles without serious investment. This method has been applied with success at Aerospatiale-Matra on various aircraft assemblies.