This work covers the problems encountered in correctly determining mode I interlaminar fracture toughness of composite materials. Pure mode I tests were performed on double cantilever beam (DCB) specimens composed by quasi-homogeneous and uncoupled multidirectional (MD) laminates using 16 or 26-ply: [α/−α2/α/−α/α2/−α]sym or anti-sym and [0/α/−α/02/−α/0/α/02/α/−α/0]sym, with α = 0°,15°,30°,45°,90°. A finite element analysis shows that the non-uniformity ratio β=(GImax−GIav.)/GIav% depends not only on the parameter Dc = D122/(D11D22), but also on the specimen geometrical ratios a/b and a/h. The condition of Dc<0.25 is not sufficient to assure a uniform GI width-wise distribution. If we want to study the crack growth between any ply angles, it is difficult to find lay-ups having β<10%. In fact, the crack initiation in MD DCB specimens usually occurred at the middle of the specimens, where GI attained a maximum. Hence, the critical energy release rate GIC has to be measured by the maximum instead of the mean of GI.