"The optimal method of measuring LA size in Mitral Regurgitation (MR) has not been determined" (Bonagura, 2008). But the LA/Ao ratio is presently the most simple approach of MR in MVD : it's an important criterion of ISACHC classification (discrimination class 1a-1b) and it correlates with many other parameters (Murmur, Heart Rate variability, LVIDd, PISA, RJA, NT proANP, ANP, NT proBNP, BNP, NO); then, it's currently used, with 2 predominant methods : the "American" one(Rishniw, 2000), and the "Swedish" one (Hansson, 2002). But exactly as in the past with the M mode approach, we can easily demonstrate the 2 main shortcomings of this method: 1) a disrespect of the physical rules of ultrasounds in their measurement instructions. 2) a disrespect in the anatomy of dilated LA. Then, both underestimate the real dimensions of LA, and the more dilated is the LA, the more important is the underestimation. What is the interest and reliability of an underestimated index? Then, we propose an adaptation of these methods, using the same small axis approach, with quite similar criterions for Ao measurement, but a LA measurement realised not on the same frozen trans-aortic image, but on a more cranial (supra-aortic) view. To assess normal LA size, 291 normal dogs from 32 different breeds were investigated, after control of the aspect of the mitral dome in long axis, and the lack of mitral leak with CFM (as defined in the French breeding program for CKC).The threshold is closed to the American method ( < 1,6) with 1,7, but in case of dilation the estimation of the real dimensions of LA is much more accurate: AG/Ao in Class 1a : 1,54 (1,17-1,88) ; Class 1b : 2,04 (1,42-3,47) ; Class 2 : 2,34 (1,94-3,46) ; Class 3a : 2,62 (1,69-3,71).